ext3-extents-2.6.18-vanilla.patch

非常经典的一个分布式系统

+	struct ext3_extent_header *eh;
+	struct buffer_head *bh;
+	short int depth, i, ppos = 0, alloc = 0;
+
+	eh = ext_inode_hdr(inode);
+	i = depth = ext_depth(inode);
+	if (ext3_ext_check_header(inode, eh, depth))
+		return ERR_PTR(-EIO);
+
+	/* account possible depth increase */
+	if (!path) {
+		path = kmalloc(sizeof(struct ext3_ext_path) * (depth + 2),
+				GFP_NOFS);
+		if (!path)
+			return ERR_PTR(-ENOMEM);
+		alloc = 1;
+	}
+	memset(path, 0, sizeof(struct ext3_ext_path) * (depth + 1));
+	path[0].p_hdr = eh;
+
+	/* walk through the tree */
+	while (i) {
+		ext_debug(inode, "depth %d: num %d, max %d\n",
+			  ppos, le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
+
+		ext3_ext_binsearch_idx(inode, path + ppos, block);
+		path[ppos].p_block = le32_to_cpu(path[ppos].p_idx->ei_leaf);
+		path[ppos].p_depth = i;
+		path[ppos].p_ext = NULL;
+
+		bh = sb_bread(inode->i_sb, path[ppos].p_block);
+		if (!bh)
+			goto err;
+
+		eh = ext_block_hdr(bh);
+		ppos++;
+		BUG_ON(ppos > depth);
+		path[ppos].p_bh = bh;
+		path[ppos].p_hdr = eh;
+		i--;
+
+		if (ext3_ext_check_header(inode, eh, i))
+			goto err;
+	}
+
+	path[ppos].p_depth = i;
+	path[ppos].p_hdr = eh;
+	path[ppos].p_ext = NULL;
+	path[ppos].p_idx = NULL;
+
+	/* find extent */
+	ext3_ext_binsearch(inode, path + ppos, block);
+
+	ext3_ext_show_path(inode, path);
+
+	return path;
+
+err:
+	ext3_ext_drop_refs(path);
+	if (alloc)
+		kfree(path);
+	return ERR_PTR(-EIO);
+}
+
+/*
+ * insert new index [logical;ptr] into the block at cupr
+ * it check where to insert: before curp or after curp
+ */
+static int ext3_ext_insert_index(handle_t *handle, struct inode *inode,
+				struct ext3_ext_path *curp,
+				int logical, int ptr)
+{
+	struct ext3_extent_idx *ix;
+	int len, err;
+
+	if ((err = ext3_ext_get_access(handle, inode, curp)))
+		return err;
+
+	BUG_ON(logical == le32_to_cpu(curp->p_idx->ei_block));
+	len = EXT_MAX_INDEX(curp->p_hdr) - curp->p_idx;
+	if (logical > le32_to_cpu(curp->p_idx->ei_block)) {
+		/* insert after */
+		if (curp->p_idx != EXT_LAST_INDEX(curp->p_hdr)) {
+			len = (len - 1) * sizeof(struct ext3_extent_idx);
+			len = len < 0 ? 0 : len;
+			ext_debug(inode, "insert new index %d after: %d. "
+					"move %d from 0x%p to 0x%p\n",
+					logical, ptr, len,
+					(curp->p_idx + 1), (curp->p_idx + 2));
+			memmove(curp->p_idx + 2, curp->p_idx + 1, len);
+		}
+		ix = curp->p_idx + 1;
+	} else {
+		/* insert before */
+		len = len * sizeof(struct ext3_extent_idx);
+		len = len < 0 ? 0 : len;
+		ext_debug(inode, "insert new index %d before: %d. "
+				"move %d from 0x%p to 0x%p\n",
+				logical, ptr, len,
+				curp->p_idx, (curp->p_idx + 1));
+		memmove(curp->p_idx + 1, curp->p_idx, len);
+		ix = curp->p_idx;
+	}
+
+	ix->ei_block = cpu_to_le32(logical);
+	ix->ei_leaf = cpu_to_le32(ptr);
+	ix->ei_leaf_hi = ix->ei_unused = 0;
+	curp->p_hdr->eh_entries = cpu_to_le16(le16_to_cpu(curp->p_hdr->eh_entries)+1);
+
+	BUG_ON(le16_to_cpu(curp->p_hdr->eh_entries)
+	                     > le16_to_cpu(curp->p_hdr->eh_max));
+	BUG_ON(ix > EXT_LAST_INDEX(curp->p_hdr));
+
+	err = ext3_ext_dirty(handle, inode, curp);
+	ext3_std_error(inode->i_sb, err);
+
+	return err;
+}
+
+/*
+ * routine inserts new subtree into the path, using free index entry
+ * at depth 'at:
+ *  - allocates all needed blocks (new leaf and all intermediate index blocks)
+ *  - makes decision where to split
+ *  - moves remaining extens and index entries (right to the split point)
+ *    into the newly allocated blocks
+ *  - initialize subtree
+ */
+static int ext3_ext_split(handle_t *handle, struct inode *inode,
+				struct ext3_ext_path *path,
+				struct ext3_extent *newext, int at)
+{
+	struct buffer_head *bh = NULL;
+	int depth = ext_depth(inode);
+	struct ext3_extent_header *neh;
+	struct ext3_extent_idx *fidx;
+	struct ext3_extent *ex;
+	int i = at, k, m, a;
+	unsigned long newblock, oldblock;
+	__le32 border;
+	int *ablocks = NULL; /* array of allocated blocks */
+	int err = 0;
+
+	/* make decision: where to split? */
+	/* FIXME: now desicion is simplest: at current extent */
+
+	/* if current leaf will be splitted, then we should use
+	 * border from split point */
+	BUG_ON(path[depth].p_ext > EXT_MAX_EXTENT(path[depth].p_hdr));
+	if (path[depth].p_ext != EXT_MAX_EXTENT(path[depth].p_hdr)) {
+		border = path[depth].p_ext[1].ee_block;
+		ext_debug(inode, "leaf will be splitted."
+				" next leaf starts at %d\n",
+			          le32_to_cpu(border));
+	} else {
+		border = newext->ee_block;
+		ext_debug(inode, "leaf will be added."
+				" next leaf starts at %d\n",
+			        le32_to_cpu(border));
+	}
+
+	/*
+	 * if error occurs, then we break processing
+	 * and turn filesystem read-only. so, index won't
+	 * be inserted and tree will be in consistent
+	 * state. next mount will repair buffers too
+	 */
+
+	/*
+	 * get array to track all allocated blocks
+	 * we need this to handle errors and free blocks
+	 * upon them
+	 */
+	ablocks = kmalloc(sizeof(unsigned long) * depth, GFP_NOFS);
+	if (!ablocks)
+		return -ENOMEM;
+	memset(ablocks, 0, sizeof(unsigned long) * depth);
+
+	/* allocate all needed blocks */
+	ext_debug(inode, "allocate %d blocks for indexes/leaf\n", depth - at);
+	for (a = 0; a < depth - at; a++) {
+		newblock = ext3_ext_new_block(handle, inode, path, newext, &err);
+		if (newblock == 0)
+			goto cleanup;
+		ablocks[a] = newblock;
+	}
+
+	/* initialize new leaf */
+	newblock = ablocks[--a];
+	BUG_ON(newblock == 0);
+	bh = sb_getblk(inode->i_sb, newblock);
+	if (!bh) {
+		err = -EIO;
+		goto cleanup;
+	}
+	lock_buffer(bh);
+
+	if ((err = ext3_journal_get_create_access(handle, bh)))
+		goto cleanup;
+
+	neh = ext_block_hdr(bh);
+	neh->eh_entries = 0;
+	neh->eh_max = cpu_to_le16(ext3_ext_space_block(inode));
+	neh->eh_magic = cpu_to_le16(EXT3_EXT_MAGIC);
+	neh->eh_depth = 0;
+	ex = EXT_FIRST_EXTENT(neh);
+
+	/* move remain of path[depth] to the new leaf */
+	BUG_ON(path[depth].p_hdr->eh_entries != path[depth].p_hdr->eh_max);
+	/* start copy from next extent */
+	/* TODO: we could do it by single memmove */
+	m = 0;
+	path[depth].p_ext++;
+	while (path[depth].p_ext <=
+			EXT_MAX_EXTENT(path[depth].p_hdr)) {
+		ext_debug(inode, "move %d:%d:%d in new leaf %lu\n",
+			        le32_to_cpu(path[depth].p_ext->ee_block),
+			        le32_to_cpu(path[depth].p_ext->ee_start),
+			        le16_to_cpu(path[depth].p_ext->ee_len),
+				newblock);
+		/*memmove(ex++, path[depth].p_ext++,
+				sizeof(struct ext3_extent));
+		neh->eh_entries++;*/
+		path[depth].p_ext++;
+		m++;
+	}
+	if (m) {
+		memmove(ex, path[depth].p_ext-m, sizeof(struct ext3_extent)*m);
+		neh->eh_entries = cpu_to_le16(le16_to_cpu(neh->eh_entries)+m);
+	}
+
+	set_buffer_uptodate(bh);
+	unlock_buffer(bh);
+
+	if ((err = ext3_journal_dirty_metadata(handle, bh)))
+		goto cleanup;
+	brelse(bh);
+	bh = NULL;
+
+	/* correct old leaf */
+	if (m) {
+		if ((err = ext3_ext_get_access(handle, inode, path + depth)))
+			goto cleanup;
+		path[depth].p_hdr->eh_entries =
+		     cpu_to_le16(le16_to_cpu(path[depth].p_hdr->eh_entries)-m);
+		if ((err = ext3_ext_dirty(handle, inode, path + depth)))
+			goto cleanup;
+
+	}
+
+	/* create intermediate indexes */
+	k = depth - at - 1;
+	BUG_ON(k < 0);
+	if (k)
+		ext_debug(inode, "create %d intermediate indices\n", k);
+	/* insert new index into current index block */
+	/* current depth stored in i var */
+	i = depth - 1;
+	while (k--) {
+		oldblock = newblock;
+		newblock = ablocks[--a];
+		bh = sb_getblk(inode->i_sb, newblock);
+		if (!bh) {
+			err = -EIO;
+			goto cleanup;
+		}
+		lock_buffer(bh);
+
+		if ((err = ext3_journal_get_create_access(handle, bh)))
+			goto cleanup;
+
+		neh = ext_block_hdr(bh);
+		neh->eh_entries = cpu_to_le16(1);
+		neh->eh_magic = cpu_to_le16(EXT3_EXT_MAGIC);
+		neh->eh_max = cpu_to_le16(ext3_ext_space_block_idx(inode));
+		neh->eh_depth = cpu_to_le16(depth - i);
+		fidx = EXT_FIRST_INDEX(neh);
+		fidx->ei_block = border;
+		fidx->ei_leaf = cpu_to_le32(oldblock);
+		fidx->ei_leaf_hi = fidx->ei_unused = 0;
+
+		ext_debug(inode, "int.index at %d (block %lu): %lu -> %lu\n", i,
+				newblock, (unsigned long) le32_to_cpu(border),
+				oldblock);
+		/* copy indexes */
+		m = 0;
+		path[i].p_idx++;
+
+		ext_debug(inode, "cur 0x%p, last 0x%p\n", path[i].p_idx,
+				EXT_MAX_INDEX(path[i].p_hdr));
+		BUG_ON(EXT_MAX_INDEX(path[i].p_hdr) !=
+				EXT_LAST_INDEX(path[i].p_hdr));
+		while (path[i].p_idx <= EXT_MAX_INDEX(path[i].p_hdr)) {
+			ext_debug(inode, "%d: move %d:%d in new index %lu\n", i,
+				        le32_to_cpu(path[i].p_idx->ei_block),
+				        le32_to_cpu(path[i].p_idx->ei_leaf),
+				        newblock);
+			/*memmove(++fidx, path[i].p_idx++,
+					sizeof(struct ext3_extent_idx));
+			neh->eh_entries++;
+			BUG_ON(neh->eh_entries > neh->eh_max);*/
+			path[i].p_idx++;
+			m++;
+		}
+		if (m) {
+			memmove(++fidx, path[i].p_idx - m,
+				sizeof(struct ext3_extent_idx) * m);
+			neh->eh_entries =
+				cpu_to_le16(le16_to_cpu(neh->eh_entries) + m);
+		}
+		set_buffer_uptodate(bh);
+		unlock_buffer(bh);
+
+		if ((err = ext3_journal_dirty_metadata(handle, bh)))
+			goto cleanup;
+		brelse(bh);
+		bh = NULL;
+
+		/* correct old index */
+		if (m) {
+			err = ext3_ext_get_access(handle, inode, path + i);
+			if (err)
+				goto cleanup;
+			path[i].p_hdr->eh_entries = cpu_to_le16(le16_to_cpu(path[i].p_hdr->eh_entries)-m);
+			err = ext3_ext_dirty(handle, inode, path + i);
+			if (err)
+				goto cleanup;
+		}
+
+		i--;
+	}
+
+	/* insert new index */
+	if (err)
+		goto cleanup;
+
+	err = ext3_ext_insert_index(handle, inode, path + at,
+				    le32_to_cpu(border), newblock);
+
+cleanup:
+	if (bh) {
+		if (buffer_locked(bh))
+			unlock_buffer(bh);
+		brelse(bh);
+	}
+
+	if (err) {
+		/* free all allocated blocks in error case */
+		for (i = 0; i < depth; i++) {
+			if (!ablocks[i])
+				continue;
+			ext3_free_blocks(handle, inode, ablocks[i], 1);
+		}
+	}
+	kfree(ablocks);
+
+	return err;
+}
+
+/*
+ * routine implements tree growing procedure:
+ *  - allocates new block
+ *  - moves top-level data (index block or leaf) into the new block
+ *  - initialize new top-level, creating index that points to the
+ *    just created block
+ */
+static int ext3_ext_grow_indepth(handle_t *handle, struct inode *inode,
+					struct ext3_ext_path *path,
+					struct ext3_extent *newext)
+{
+	struct ext3_ext_path *curp = path;
+	struct ext3_extent_header *neh;
+	struct ext3_extent_idx *fidx;
+	struct buffer_head *bh;
+	unsigned long newblock;
+	int err = 0;
+
+	newblock = ext3_ext_new_block(handle, inode, path, newext, &err);
+	if (newblock == 0)
+		return err;
+
+	bh = sb_getblk(inode->i_sb, newblock);
+	if (!bh) {
+		err = -EIO;
+		ext3_std_error(inode->i_sb, err);
+		return err;
+	}
+	lock_buffer(bh);
+
+	if ((err = ext3_journal_get_create_access(handle, bh))) {
+		unlock_buffer(bh);
+		goto out;
+	}
+
+	/* move top-level index/leaf into new block */
+	memmove(bh->b_data, curp->p_hdr, sizeof(EXT3_I(inode)->i_data));
+
+	/* set size of new block */
+	neh = ext_block_hdr(bh);
+	/* old root could have indexes or leaves
+	 * so calculate e_max right way */
+	if (ext_depth(inode))
+	  neh->eh_max = cpu_to_le16(ext3_ext_space_block_idx(inode));
+	else
+	  neh->eh_max = cpu_to_le16(ext3_ext_space_block(inode));
+	neh->eh_magic = cpu_to_le16(EXT3_EXT_MAGIC);
+	set_buffer_uptodate(bh);
+	unlock_buffer(bh);
+
+	if ((err = ext3_journal_dirty_metadata(handle, bh)))
+		goto out;
+
+	/* create index in new top-level index: num,max,pointer */
+	if ((err = ext3_ext_get_access(handle, inode, curp)))
+		goto out;
+
+	curp->p_hdr->eh_magic = cpu_to_le16(EXT3_EXT_MAGIC);
+	curp->p_hdr->eh_max = cpu_to_le16(ext3_ext_space_root_idx(inode));
+	curp->p_hdr->eh_entries = cpu_to_le16(1);
+	curp->p_idx = EXT_FIRST_INDEX(curp->p_hdr);
+	/* FIXME: it works, but actually path[0] can be index */
+	curp->p_idx->ei_block = EXT_FIRST_EXTENT(path[0].p_hdr)->ee_block;
+	curp->p_idx->ei_leaf = cpu_to_le32(newblock);
+	curp->p_idx->ei_leaf_hi = curp->p_idx->ei_unused = 0;
+
+	neh = ext_inode_hdr(inode);
+	fidx = EXT_FIRST_INDEX(neh);
+	ext_debug(inode, "new root: num %d(%d), lblock %d, ptr %d\n",
+		  le16_to_cpu(neh->eh_entries), le16_to_cpu(neh->eh_max),
+		  le32_to_cpu(fidx->ei_block), le32_to_cpu(fidx->ei_leaf));
+
+	neh->eh_depth = cpu_to_le16(path->p_depth + 1);
+	err = ext3_ext_dirty(handle, inode, curp);
+out:
+	brelse(bh);
+
+	return err;
+}
+
+/*
+ * routine finds empty index and adds new leaf. if no free index found
+ * then it requests in-depth growing
+ */
+static int ext3_ext_create_new_leaf(handle_t *handle, struct inode *inode,
+					struct ext3_ext_path *path,
+					struct ext3_extent *newext)
+{
+	struct ext3_ext_path *curp;
+	int depth, i, err = 0;
+
+repeat:
+	i = depth = ext_depth(inode);
+
+	/* walk up to the tree and look for free index entry */
+	curp = path + depth;
+	while (i > 0 && !EXT_HAS_FREE_INDEX(curp)) {
+		i--;
+		curp--;
+	}
+
+	/* we use already allocated block for index block
+	 * so, subsequent data blocks should be contigoues */
+	if (EXT_HAS_FREE_INDEX(curp)) {
+		/* if we found index with free entry, then use that
+		 * entry: create all needed subtree and add new leaf */
+		err = ext3_ext_split(handle, inode, path, newext, i);
+		if (err)
+			goto out;
+
+		/* refill path */
+		ext3_ext_drop_refs(path);