[linux-2.6-block.git] / fs / hfs / bnode.c

/*
 *  linux/fs/hfs/bnode.c
 *
 * Copyright (C) 2001
 * Brad Boyer (flar@allandria.com)
 * (C) 2003 Ardis Technologies <roman@ardistech.com>
 *
 * Handle basic btree node operations
 */

#include <linux/pagemap.h>
#include <linux/slab.h>
#include <linux/swap.h>

#include "btree.h"

void hfs_bnode_read(struct hfs_bnode *node, void *buf,
		int off, int len)
{
	struct page *page;

	off += node->page_offset;
	page = node->page[0];

	memcpy(buf, kmap(page) + off, len);
	kunmap(page);
}

u16 hfs_bnode_read_u16(struct hfs_bnode *node, int off)
{
	__be16 data;
	// optimize later...
	hfs_bnode_read(node, &data, off, 2);
	return be16_to_cpu(data);
}

u8 hfs_bnode_read_u8(struct hfs_bnode *node, int off)
{
	u8 data;
	// optimize later...
	hfs_bnode_read(node, &data, off, 1);
	return data;
}

void hfs_bnode_read_key(struct hfs_bnode *node, void *key, int off)
{
	struct hfs_btree *tree;
	int key_len;

	tree = node->tree;
	if (node->type == HFS_NODE_LEAF ||
	    tree->attributes & HFS_TREE_VARIDXKEYS)
		key_len = hfs_bnode_read_u8(node, off) + 1;
	else
		key_len = tree->max_key_len + 1;

	hfs_bnode_read(node, key, off, key_len);
}

void hfs_bnode_write(struct hfs_bnode *node, void *buf, int off, int len)
{
	struct page *page;

	off += node->page_offset;
	page = node->page[0];

	memcpy(kmap(page) + off, buf, len);
	kunmap(page);
	set_page_dirty(page);
}

void hfs_bnode_write_u16(struct hfs_bnode *node, int off, u16 data)
{
	__be16 v = cpu_to_be16(data);
	// optimize later...
	hfs_bnode_write(node, &v, off, 2);
}

void hfs_bnode_write_u8(struct hfs_bnode *node, int off, u8 data)
{
	// optimize later...
	hfs_bnode_write(node, &data, off, 1);
}

void hfs_bnode_clear(struct hfs_bnode *node, int off, int len)
{
	struct page *page;

	off += node->page_offset;
	page = node->page[0];

	memset(kmap(page) + off, 0, len);
	kunmap(page);
	set_page_dirty(page);
}

void hfs_bnode_copy(struct hfs_bnode *dst_node, int dst,
		struct hfs_bnode *src_node, int src, int len)
{
	struct hfs_btree *tree;
	struct page *src_page, *dst_page;

	hfs_dbg(BNODE_MOD, "copybytes: %u,%u,%u\n", dst, src, len);
	if (!len)
		return;
	tree = src_node->tree;
	src += src_node->page_offset;
	dst += dst_node->page_offset;
	src_page = src_node->page[0];
	dst_page = dst_node->page[0];

	memcpy(kmap(dst_page) + dst, kmap(src_page) + src, len);
	kunmap(src_page);
	kunmap(dst_page);
	set_page_dirty(dst_page);
}

void hfs_bnode_move(struct hfs_bnode *node, int dst, int src, int len)
{
	struct page *page;
	void *ptr;

	hfs_dbg(BNODE_MOD, "movebytes: %u,%u,%u\n", dst, src, len);
	if (!len)
		return;
	src += node->page_offset;
	dst += node->page_offset;
	page = node->page[0];
	ptr = kmap(page);
	memmove(ptr + dst, ptr + src, len);
	kunmap(page);
	set_page_dirty(page);
}

void hfs_bnode_dump(struct hfs_bnode *node)
{
	struct hfs_bnode_desc desc;
	__be32 cnid;
	int i, off, key_off;

	hfs_dbg(BNODE_MOD, "bnode: %d\n", node->this);
	hfs_bnode_read(node, &desc, 0, sizeof(desc));
	hfs_dbg(BNODE_MOD, "%d, %d, %d, %d, %d\n",
		be32_to_cpu(desc.next), be32_to_cpu(desc.prev),
		desc.type, desc.height, be16_to_cpu(desc.num_recs));

	off = node->tree->node_size - 2;
	for (i = be16_to_cpu(desc.num_recs); i >= 0; off -= 2, i--) {
		key_off = hfs_bnode_read_u16(node, off);
		hfs_dbg_cont(BNODE_MOD, " %d", key_off);
		if (i && node->type == HFS_NODE_INDEX) {
			int tmp;

			if (node->tree->attributes & HFS_TREE_VARIDXKEYS)
				tmp = (hfs_bnode_read_u8(node, key_off) | 1) + 1;
			else
				tmp = node->tree->max_key_len + 1;
			hfs_dbg_cont(BNODE_MOD, " (%d,%d",
				     tmp, hfs_bnode_read_u8(node, key_off));
			hfs_bnode_read(node, &cnid, key_off + tmp, 4);
			hfs_dbg_cont(BNODE_MOD, ",%d)", be32_to_cpu(cnid));
		} else if (i && node->type == HFS_NODE_LEAF) {
			int tmp;

			tmp = hfs_bnode_read_u8(node, key_off);
			hfs_dbg_cont(BNODE_MOD, " (%d)", tmp);
		}
	}
	hfs_dbg_cont(BNODE_MOD, "\n");
}

void hfs_bnode_unlink(struct hfs_bnode *node)
{
	struct hfs_btree *tree;
	struct hfs_bnode *tmp;
	__be32 cnid;

	tree = node->tree;
	if (node->prev) {
		tmp = hfs_bnode_find(tree, node->prev);
		if (IS_ERR(tmp))
			return;
		tmp->next = node->next;
		cnid = cpu_to_be32(tmp->next);
		hfs_bnode_write(tmp, &cnid, offsetof(struct hfs_bnode_desc, next), 4);
		hfs_bnode_put(tmp);
	} else if (node->type == HFS_NODE_LEAF)
		tree->leaf_head = node->next;

	if (node->next) {
		tmp = hfs_bnode_find(tree, node->next);
		if (IS_ERR(tmp))
			return;
		tmp->prev = node->prev;
		cnid = cpu_to_be32(tmp->prev);
		hfs_bnode_write(tmp, &cnid, offsetof(struct hfs_bnode_desc, prev), 4);
		hfs_bnode_put(tmp);
	} else if (node->type == HFS_NODE_LEAF)
		tree->leaf_tail = node->prev;

	// move down?
	if (!node->prev && !node->next) {
		printk(KERN_DEBUG "hfs_btree_del_level\n");
	}
	if (!node->parent) {
		tree->root = 0;
		tree->depth = 0;
	}
	set_bit(HFS_BNODE_DELETED, &node->flags);
}

static inline int hfs_bnode_hash(u32 num)
{
	num = (num >> 16) + num;
	num += num >> 8;
	return num & (NODE_HASH_SIZE - 1);
}

struct hfs_bnode *hfs_bnode_findhash(struct hfs_btree *tree, u32 cnid)
{
	struct hfs_bnode *node;

	if (cnid >= tree->node_count) {
		pr_err("request for non-existent node %d in B*Tree\n", cnid);
		return NULL;
	}

	for (node = tree->node_hash[hfs_bnode_hash(cnid)];
	     node; node = node->next_hash) {
		if (node->this == cnid) {
			return node;
		}
	}
	return NULL;
}

static struct hfs_bnode *__hfs_bnode_create(struct hfs_btree *tree, u32 cnid)
{
	struct super_block *sb;
	struct hfs_bnode *node, *node2;
	struct address_space *mapping;
	struct page *page;
	int size, block, i, hash;
	loff_t off;

	if (cnid >= tree->node_count) {
		pr_err("request for non-existent node %d in B*Tree\n", cnid);
		return NULL;
	}

	sb = tree->inode->i_sb;
	size = sizeof(struct hfs_bnode) + tree->pages_per_bnode *
		sizeof(struct page *);
	node = kzalloc(size, GFP_KERNEL);
	if (!node)
		return NULL;
	node->tree = tree;
	node->this = cnid;
	set_bit(HFS_BNODE_NEW, &node->flags);
	atomic_set(&node->refcnt, 1);
	hfs_dbg(BNODE_REFS, "new_node(%d:%d): 1\n",
		node->tree->cnid, node->this);
	init_waitqueue_head(&node->lock_wq);
	spin_lock(&tree->hash_lock);
	node2 = hfs_bnode_findhash(tree, cnid);
	if (!node2) {
		hash = hfs_bnode_hash(cnid);
		node->next_hash = tree->node_hash[hash];
		tree->node_hash[hash] = node;
		tree->node_hash_cnt++;
	} else {
		spin_unlock(&tree->hash_lock);
		kfree(node);
		wait_event(node2->lock_wq, !test_bit(HFS_BNODE_NEW, &node2->flags));
		return node2;
	}
	spin_unlock(&tree->hash_lock);

	mapping = tree->inode->i_mapping;
	off = (loff_t)cnid * tree->node_size;
	block = off >> PAGE_SHIFT;
	node->page_offset = off & ~PAGE_MASK;
	for (i = 0; i < tree->pages_per_bnode; i++) {
		page = read_mapping_page(mapping, block++, NULL);
		if (IS_ERR(page))
			goto fail;
		if (PageError(page)) {
			put_page(page);
			goto fail;
		}
		node->page[i] = page;
	}

	return node;
fail:
	set_bit(HFS_BNODE_ERROR, &node->flags);
	return node;
}

void hfs_bnode_unhash(struct hfs_bnode *node)
{
	struct hfs_bnode **p;

	hfs_dbg(BNODE_REFS, "remove_node(%d:%d): %d\n",
		node->tree->cnid, node->this, atomic_read(&node->refcnt));
	for (p = &node->tree->node_hash[hfs_bnode_hash(node->this)];
	     *p && *p != node; p = &(*p)->next_hash)
		;
	BUG_ON(!*p);
	*p = node->next_hash;
	node->tree->node_hash_cnt--;
}

/* Load a particular node out of a tree */
struct hfs_bnode *hfs_bnode_find(struct hfs_btree *tree, u32 num)
{
	struct hfs_bnode *node;
	struct hfs_bnode_desc *desc;
	int i, rec_off, off, next_off;
	int entry_size, key_size;

	spin_lock(&tree->hash_lock);
	node = hfs_bnode_findhash(tree, num);
	if (node) {
		hfs_bnode_get(node);
		spin_unlock(&tree->hash_lock);
		wait_event(node->lock_wq, !test_bit(HFS_BNODE_NEW, &node->flags));
		if (test_bit(HFS_BNODE_ERROR, &node->flags))
			goto node_error;
		return node;
	}
	spin_unlock(&tree->hash_lock);
	node = __hfs_bnode_create(tree, num);
	if (!node)
		return ERR_PTR(-ENOMEM);
	if (test_bit(HFS_BNODE_ERROR, &node->flags))
		goto node_error;
	if (!test_bit(HFS_BNODE_NEW, &node->flags))
		return node;

	desc = (struct hfs_bnode_desc *)(kmap(node->page[0]) + node->page_offset);
	node->prev = be32_to_cpu(desc->prev);
	node->next = be32_to_cpu(desc->next);
	node->num_recs = be16_to_cpu(desc->num_recs);
	node->type = desc->type;
	node->height = desc->height;
	kunmap(node->page[0]);

	switch (node->type) {
	case HFS_NODE_HEADER:
	case HFS_NODE_MAP:
		if (node->height != 0)
			goto node_error;
		break;
	case HFS_NODE_LEAF:
		if (node->height != 1)
			goto node_error;
		break;
	case HFS_NODE_INDEX:
		if (node->height <= 1 || node->height > tree->depth)
			goto node_error;
		break;
	default:
		goto node_error;
	}

	rec_off = tree->node_size - 2;
	off = hfs_bnode_read_u16(node, rec_off);
	if (off != sizeof(struct hfs_bnode_desc))
		goto node_error;
	for (i = 1; i <= node->num_recs; off = next_off, i++) {
		rec_off -= 2;
		next_off = hfs_bnode_read_u16(node, rec_off);
		if (next_off <= off ||
		    next_off > tree->node_size ||
		    next_off & 1)
			goto node_error;
		entry_size = next_off - off;
		if (node->type != HFS_NODE_INDEX &&
		    node->type != HFS_NODE_LEAF)
			continue;
		key_size = hfs_bnode_read_u8(node, off) + 1;
		if (key_size >= entry_size /*|| key_size & 1*/)
			goto node_error;
	}
	clear_bit(HFS_BNODE_NEW, &node->flags);
	wake_up(&node->lock_wq);
	return node;

node_error:
	set_bit(HFS_BNODE_ERROR, &node->flags);
	clear_bit(HFS_BNODE_NEW, &node->flags);
	wake_up(&node->lock_wq);
	hfs_bnode_put(node);
	return ERR_PTR(-EIO);
}

void hfs_bnode_free(struct hfs_bnode *node)
{
	int i;

	for (i = 0; i < node->tree->pages_per_bnode; i++)
		if (node->page[i])
			put_page(node->page[i]);
	kfree(node);
}

struct hfs_bnode *hfs_bnode_create(struct hfs_btree *tree, u32 num)
{
	struct hfs_bnode *node;
	struct page **pagep;
	int i;

	spin_lock(&tree->hash_lock);
	node = hfs_bnode_findhash(tree, num);
	spin_unlock(&tree->hash_lock);
	if (node) {
		pr_crit("new node %u already hashed?\n", num);
		WARN_ON(1);
		return node;
	}
	node = __hfs_bnode_create(tree, num);
	if (!node)
		return ERR_PTR(-ENOMEM);
	if (test_bit(HFS_BNODE_ERROR, &node->flags)) {
		hfs_bnode_put(node);
		return ERR_PTR(-EIO);
	}

	pagep = node->page;
	memset(kmap(*pagep) + node->page_offset, 0,
	       min((int)PAGE_SIZE, (int)tree->node_size));
	set_page_dirty(*pagep);
	kunmap(*pagep);
	for (i = 1; i < tree->pages_per_bnode; i++) {
		memset(kmap(*++pagep), 0, PAGE_SIZE);
		set_page_dirty(*pagep);
		kunmap(*pagep);
	}
	clear_bit(HFS_BNODE_NEW, &node->flags);
	wake_up(&node->lock_wq);

	return node;
}

void hfs_bnode_get(struct hfs_bnode *node)
{
	if (node) {
		atomic_inc(&node->refcnt);
		hfs_dbg(BNODE_REFS, "get_node(%d:%d): %d\n",
			node->tree->cnid, node->this,
			atomic_read(&node->refcnt));
	}
}

/* Dispose of resources used by a node */
void hfs_bnode_put(struct hfs_bnode *node)
{
	if (node) {
		struct hfs_btree *tree = node->tree;
		int i;

		hfs_dbg(BNODE_REFS, "put_node(%d:%d): %d\n",
			node->tree->cnid, node->this,
			atomic_read(&node->refcnt));
		BUG_ON(!atomic_read(&node->refcnt));
		if (!atomic_dec_and_lock(&node->refcnt, &tree->hash_lock))
			return;
		for (i = 0; i < tree->pages_per_bnode; i++) {
			if (!node->page[i])
				continue;
			mark_page_accessed(node->page[i]);
		}

		if (test_bit(HFS_BNODE_DELETED, &node->flags)) {
			hfs_bnode_unhash(node);
			spin_unlock(&tree->hash_lock);
			hfs_bmap_free(node);
			hfs_bnode_free(node);
			return;
		}
		spin_unlock(&tree->hash_lock);
	}
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/hfs/bnode.c
	3	*
	4	* Copyright (C) 2001
	5	* Brad Boyer (flar@allandria.com)
	6	* (C) 2003 Ardis Technologies <roman@ardistech.com>
	7	*
	8	* Handle basic btree node operations
	9	*/
	10
	11	#include <linux/pagemap.h>
5a0e3ad6	12	#include <linux/slab.h>
1da177e4 LT	13	#include <linux/swap.h>
	14
	15	#include "btree.h"
	16
1da177e4 LT	17	void hfs_bnode_read(struct hfs_bnode node, void buf,
	18	int off, int len)
	19	{
	20	struct page *page;
	21
	22	off += node->page_offset;
	23	page = node->page[0];
	24
	25	memcpy(buf, kmap(page) + off, len);
	26	kunmap(page);
	27	}
	28
	29	u16 hfs_bnode_read_u16(struct hfs_bnode *node, int off)
	30	{
	31	__be16 data;
	32	// optimize later...
	33	hfs_bnode_read(node, &data, off, 2);
	34	return be16_to_cpu(data);
	35	}
	36
	37	u8 hfs_bnode_read_u8(struct hfs_bnode *node, int off)
	38	{
	39	u8 data;
	40	// optimize later...
	41	hfs_bnode_read(node, &data, off, 1);
	42	return data;
	43	}
	44
	45	void hfs_bnode_read_key(struct hfs_bnode node, void key, int off)
	46	{
	47	struct hfs_btree *tree;
	48	int key_len;
	49
	50	tree = node->tree;
	51	if (node->type == HFS_NODE_LEAF \|\|
	52	tree->attributes & HFS_TREE_VARIDXKEYS)
	53	key_len = hfs_bnode_read_u8(node, off) + 1;
	54	else
	55	key_len = tree->max_key_len + 1;
	56
	57	hfs_bnode_read(node, key, off, key_len);
	58	}
	59
	60	void hfs_bnode_write(struct hfs_bnode node, void buf, int off, int len)
	61	{
	62	struct page *page;
	63
	64	off += node->page_offset;
	65	page = node->page[0];
	66
	67	memcpy(kmap(page) + off, buf, len);
	68	kunmap(page);
	69	set_page_dirty(page);
	70	}
	71
	72	void hfs_bnode_write_u16(struct hfs_bnode *node, int off, u16 data)
	73	{
	74	__be16 v = cpu_to_be16(data);
	75	// optimize later...
	76	hfs_bnode_write(node, &v, off, 2);
	77	}
	78
	79	void hfs_bnode_write_u8(struct hfs_bnode *node, int off, u8 data)
	80	{
81	// optimize later...
82	hfs_bnode_write(node, &data, off, 1);
83	}
84
85	void hfs_bnode_clear(struct hfs_bnode *node, int off, int len)
86	{
87	struct page *page;
88
89	off += node->page_offset;
90	page = node->page[0];
91
92	memset(kmap(page) + off, 0, len);
93	kunmap(page);
94	set_page_dirty(page);
95	}
96
97	void hfs_bnode_copy(struct hfs_bnode *dst_node, int dst,
98	struct hfs_bnode *src_node, int src, int len)
99	{
100	struct hfs_btree *tree;
101	struct page src_page, dst_page;
102
c2b3e1f7	103	hfs_dbg(BNODE_MOD, "copybytes: %u,%u,%u\n", dst, src, len);
1da177e4 LT	104	if (!len)
	105	return;
	106	tree = src_node->tree;
	107	src += src_node->page_offset;
	108	dst += dst_node->page_offset;
	109	src_page = src_node->page[0];
	110	dst_page = dst_node->page[0];
	111
	112	memcpy(kmap(dst_page) + dst, kmap(src_page) + src, len);
	113	kunmap(src_page);
	114	kunmap(dst_page);
	115	set_page_dirty(dst_page);
	116	}
	117
	118	void hfs_bnode_move(struct hfs_bnode *node, int dst, int src, int len)
	119	{
	120	struct page *page;
	121	void *ptr;
	122
c2b3e1f7	123	hfs_dbg(BNODE_MOD, "movebytes: %u,%u,%u\n", dst, src, len);
1da177e4 LT	124	if (!len)
	125	return;
	126	src += node->page_offset;
	127	dst += node->page_offset;
	128	page = node->page[0];
	129	ptr = kmap(page);
	130	memmove(ptr + dst, ptr + src, len);
	131	kunmap(page);
	132	set_page_dirty(page);
	133	}
	134
	135	void hfs_bnode_dump(struct hfs_bnode *node)
	136	{
	137	struct hfs_bnode_desc desc;
	138	__be32 cnid;
	139	int i, off, key_off;
	140
c2b3e1f7	141	hfs_dbg(BNODE_MOD, "bnode: %d\n", node->this);
1da177e4	142	hfs_bnode_read(node, &desc, 0, sizeof(desc));
c2b3e1f7	143	hfs_dbg(BNODE_MOD, "%d, %d, %d, %d, %d\n",
1da177e4 LT	144	be32_to_cpu(desc.next), be32_to_cpu(desc.prev),
	145	desc.type, desc.height, be16_to_cpu(desc.num_recs));
	146
	147	off = node->tree->node_size - 2;
	148	for (i = be16_to_cpu(desc.num_recs); i >= 0; off -= 2, i--) {
	149	key_off = hfs_bnode_read_u16(node, off);
c2b3e1f7	150	hfs_dbg_cont(BNODE_MOD, " %d", key_off);
1da177e4 LT	151	if (i && node->type == HFS_NODE_INDEX) {
	152	int tmp;
	153
	154	if (node->tree->attributes & HFS_TREE_VARIDXKEYS)
	155	tmp = (hfs_bnode_read_u8(node, key_off) \| 1) + 1;
	156	else
	157	tmp = node->tree->max_key_len + 1;
c2b3e1f7 JP	158	hfs_dbg_cont(BNODE_MOD, " (%d,%d",
c2b3e1f7 JP	159	tmp, hfs_bnode_read_u8(node, key_off));
1da177e4	160	hfs_bnode_read(node, &cnid, key_off + tmp, 4);
c2b3e1f7	161	hfs_dbg_cont(BNODE_MOD, ",%d)", be32_to_cpu(cnid));
1da177e4 LT	162	} else if (i && node->type == HFS_NODE_LEAF) {
	163	int tmp;
	164
	165	tmp = hfs_bnode_read_u8(node, key_off);
c2b3e1f7	166	hfs_dbg_cont(BNODE_MOD, " (%d)", tmp);
1da177e4 LT	167	}
1da177e4 LT	168	}
c2b3e1f7	169	hfs_dbg_cont(BNODE_MOD, "\n");
1da177e4 LT	170	}
	171
	172	void hfs_bnode_unlink(struct hfs_bnode *node)
	173	{
	174	struct hfs_btree *tree;
	175	struct hfs_bnode *tmp;
	176	__be32 cnid;
	177
	178	tree = node->tree;
	179	if (node->prev) {
	180	tmp = hfs_bnode_find(tree, node->prev);
	181	if (IS_ERR(tmp))
	182	return;
	183	tmp->next = node->next;
	184	cnid = cpu_to_be32(tmp->next);
	185	hfs_bnode_write(tmp, &cnid, offsetof(struct hfs_bnode_desc, next), 4);
	186	hfs_bnode_put(tmp);
	187	} else if (node->type == HFS_NODE_LEAF)
	188	tree->leaf_head = node->next;
	189
	190	if (node->next) {
	191	tmp = hfs_bnode_find(tree, node->next);
	192	if (IS_ERR(tmp))
	193	return;
	194	tmp->prev = node->prev;
	195	cnid = cpu_to_be32(tmp->prev);
	196	hfs_bnode_write(tmp, &cnid, offsetof(struct hfs_bnode_desc, prev), 4);
	197	hfs_bnode_put(tmp);
	198	} else if (node->type == HFS_NODE_LEAF)
	199	tree->leaf_tail = node->prev;
	200
	201	// move down?
	202	if (!node->prev && !node->next) {
7cf3cc30	203	printk(KERN_DEBUG "hfs_btree_del_level\n");
1da177e4 LT	204	}
	205	if (!node->parent) {
	206	tree->root = 0;
	207	tree->depth = 0;
	208	}
	209	set_bit(HFS_BNODE_DELETED, &node->flags);
	210	}
	211
	212	static inline int hfs_bnode_hash(u32 num)
	213	{
	214	num = (num >> 16) + num;
	215	num += num >> 8;
	216	return num & (NODE_HASH_SIZE - 1);
	217	}
	218
	219	struct hfs_bnode hfs_bnode_findhash(struct hfs_btree tree, u32 cnid)
	220	{
	221	struct hfs_bnode *node;
	222
	223	if (cnid >= tree->node_count) {
d6142673	224	pr_err("request for non-existent node %d in B*Tree\n", cnid);
1da177e4 LT	225	return NULL;
	226	}
	227
	228	for (node = tree->node_hash[hfs_bnode_hash(cnid)];
	229	node; node = node->next_hash) {
	230	if (node->this == cnid) {
	231	return node;
	232	}
	233	}
	234	return NULL;
	235	}
	236
	237	static struct hfs_bnode __hfs_bnode_create(struct hfs_btree tree, u32 cnid)
	238	{
	239	struct super_block *sb;
	240	struct hfs_bnode node, node2;
	241	struct address_space *mapping;
	242	struct page *page;
	243	int size, block, i, hash;
	244	loff_t off;
	245
	246	if (cnid >= tree->node_count) {
d6142673	247	pr_err("request for non-existent node %d in B*Tree\n", cnid);
1da177e4 LT	248	return NULL;
	249	}
	250
	251	sb = tree->inode->i_sb;
	252	size = sizeof(struct hfs_bnode) + tree->pages_per_bnode *
	253	sizeof(struct page *);
f8314dc6	254	node = kzalloc(size, GFP_KERNEL);
1da177e4 LT	255	if (!node)
1da177e4 LT	256	return NULL;
1da177e4 LT	257	node->tree = tree;
	258	node->this = cnid;
	259	set_bit(HFS_BNODE_NEW, &node->flags);
	260	atomic_set(&node->refcnt, 1);
c2b3e1f7 JP	261	hfs_dbg(BNODE_REFS, "new_node(%d:%d): 1\n",
c2b3e1f7 JP	262	node->tree->cnid, node->this);
1da177e4 LT	263	init_waitqueue_head(&node->lock_wq);
	264	spin_lock(&tree->hash_lock);
	265	node2 = hfs_bnode_findhash(tree, cnid);
	266	if (!node2) {
	267	hash = hfs_bnode_hash(cnid);
	268	node->next_hash = tree->node_hash[hash];
	269	tree->node_hash[hash] = node;
	270	tree->node_hash_cnt++;
	271	} else {
	272	spin_unlock(&tree->hash_lock);
	273	kfree(node);
	274	wait_event(node2->lock_wq, !test_bit(HFS_BNODE_NEW, &node2->flags));
	275	return node2;
	276	}
	277	spin_unlock(&tree->hash_lock);
	278
	279	mapping = tree->inode->i_mapping;
	280	off = (loff_t)cnid * tree->node_size;
09cbfeaf KS	281	block = off >> PAGE_SHIFT;
09cbfeaf KS	282	node->page_offset = off & ~PAGE_MASK;
1da177e4	283	for (i = 0; i < tree->pages_per_bnode; i++) {
090d2b18	284	page = read_mapping_page(mapping, block++, NULL);
1da177e4 LT	285	if (IS_ERR(page))
	286	goto fail;
	287	if (PageError(page)) {
09cbfeaf	288	put_page(page);
1da177e4 LT	289	goto fail;
1da177e4 LT	290	}
1da177e4 LT	291	node->page[i] = page;
	292	}
	293
	294	return node;
	295	fail:
	296	set_bit(HFS_BNODE_ERROR, &node->flags);
	297	return node;
	298	}
	299
	300	void hfs_bnode_unhash(struct hfs_bnode *node)
	301	{
	302	struct hfs_bnode **p;
	303
c2b3e1f7	304	hfs_dbg(BNODE_REFS, "remove_node(%d:%d): %d\n",
1da177e4 LT	305	node->tree->cnid, node->this, atomic_read(&node->refcnt));
	306	for (p = &node->tree->node_hash[hfs_bnode_hash(node->this)];
	307	p && p != node; p = &(*p)->next_hash)
	308	;
4d4ef9ab	309	BUG_ON(!*p);
1da177e4 LT	310	*p = node->next_hash;
	311	node->tree->node_hash_cnt--;
	312	}
	313
	314	/* Load a particular node out of a tree */
	315	struct hfs_bnode hfs_bnode_find(struct hfs_btree tree, u32 num)
	316	{
	317	struct hfs_bnode *node;
	318	struct hfs_bnode_desc *desc;
	319	int i, rec_off, off, next_off;
	320	int entry_size, key_size;
	321
	322	spin_lock(&tree->hash_lock);
	323	node = hfs_bnode_findhash(tree, num);
	324	if (node) {
	325	hfs_bnode_get(node);
	326	spin_unlock(&tree->hash_lock);
	327	wait_event(node->lock_wq, !test_bit(HFS_BNODE_NEW, &node->flags));
	328	if (test_bit(HFS_BNODE_ERROR, &node->flags))
	329	goto node_error;
	330	return node;
	331	}
	332	spin_unlock(&tree->hash_lock);
	333	node = __hfs_bnode_create(tree, num);
	334	if (!node)
	335	return ERR_PTR(-ENOMEM);
	336	if (test_bit(HFS_BNODE_ERROR, &node->flags))
	337	goto node_error;
	338	if (!test_bit(HFS_BNODE_NEW, &node->flags))
	339	return node;
	340
	341	desc = (struct hfs_bnode_desc *)(kmap(node->page[0]) + node->page_offset);
	342	node->prev = be32_to_cpu(desc->prev);
	343	node->next = be32_to_cpu(desc->next);
	344	node->num_recs = be16_to_cpu(desc->num_recs);
	345	node->type = desc->type;
	346	node->height = desc->height;
	347	kunmap(node->page[0]);
	348
	349	switch (node->type) {
	350	case HFS_NODE_HEADER:
	351	case HFS_NODE_MAP:
	352	if (node->height != 0)
	353	goto node_error;
	354	break;
	355	case HFS_NODE_LEAF:
	356	if (node->height != 1)
	357	goto node_error;
	358	break;
	359	case HFS_NODE_INDEX:
	360	if (node->height <= 1 \|\| node->height > tree->depth)
	361	goto node_error;
	362	break;
	363	default:
	364	goto node_error;
	365	}
	366
	367	rec_off = tree->node_size - 2;
	368	off = hfs_bnode_read_u16(node, rec_off);
	369	if (off != sizeof(struct hfs_bnode_desc))
	370	goto node_error;
	371	for (i = 1; i <= node->num_recs; off = next_off, i++) {
	372	rec_off -= 2;
	373	next_off = hfs_bnode_read_u16(node, rec_off);
374	if (next_off <= off \|\|
375	next_off > tree->node_size \|\|
376	next_off & 1)
377	goto node_error;
378	entry_size = next_off - off;
379	if (node->type != HFS_NODE_INDEX &&
380	node->type != HFS_NODE_LEAF)
381	continue;
382	key_size = hfs_bnode_read_u8(node, off) + 1;
383	if (key_size >= entry_size /\|\| key_size & 1/)
384	goto node_error;
385	}
386	clear_bit(HFS_BNODE_NEW, &node->flags);
387	wake_up(&node->lock_wq);
388	return node;
389
390	node_error:
391	set_bit(HFS_BNODE_ERROR, &node->flags);
392	clear_bit(HFS_BNODE_NEW, &node->flags);
393	wake_up(&node->lock_wq);
394	hfs_bnode_put(node);
395	return ERR_PTR(-EIO);
396	}
397
398	void hfs_bnode_free(struct hfs_bnode *node)
399	{
7cb74be6	400	int i;
1da177e4	401
7cb74be6 HTL	402	for (i = 0; i < node->tree->pages_per_bnode; i++)
7cb74be6 HTL	403	if (node->page[i])
09cbfeaf	404	put_page(node->page[i]);
1da177e4 LT	405	kfree(node);
	406	}
	407
	408	struct hfs_bnode hfs_bnode_create(struct hfs_btree tree, u32 num)
	409	{
	410	struct hfs_bnode *node;
	411	struct page **pagep;
	412	int i;
	413
	414	spin_lock(&tree->hash_lock);
	415	node = hfs_bnode_findhash(tree, num);
	416	spin_unlock(&tree->hash_lock);
fb09c373 JM	417	if (node) {
	418	pr_crit("new node %u already hashed?\n", num);
	419	WARN_ON(1);
	420	return node;
	421	}
1da177e4 LT	422	node = __hfs_bnode_create(tree, num);
	423	if (!node)
	424	return ERR_PTR(-ENOMEM);
	425	if (test_bit(HFS_BNODE_ERROR, &node->flags)) {
	426	hfs_bnode_put(node);
	427	return ERR_PTR(-EIO);
	428	}
	429
	430	pagep = node->page;
	431	memset(kmap(*pagep) + node->page_offset, 0,
09cbfeaf	432	min((int)PAGE_SIZE, (int)tree->node_size));
1da177e4 LT	433	set_page_dirty(*pagep);
	434	kunmap(*pagep);
	435	for (i = 1; i < tree->pages_per_bnode; i++) {
09cbfeaf	436	memset(kmap(*++pagep), 0, PAGE_SIZE);
1da177e4 LT	437	set_page_dirty(*pagep);
	438	kunmap(*pagep);
	439	}
	440	clear_bit(HFS_BNODE_NEW, &node->flags);
	441	wake_up(&node->lock_wq);
	442
	443	return node;
	444	}
	445
	446	void hfs_bnode_get(struct hfs_bnode *node)
	447	{
	448	if (node) {
	449	atomic_inc(&node->refcnt);
c2b3e1f7 JP	450	hfs_dbg(BNODE_REFS, "get_node(%d:%d): %d\n",
	451	node->tree->cnid, node->this,
	452	atomic_read(&node->refcnt));
1da177e4 LT	453	}
	454	}
	455
	456	/* Dispose of resources used by a node */
	457	void hfs_bnode_put(struct hfs_bnode *node)
	458	{
	459	if (node) {
	460	struct hfs_btree *tree = node->tree;
	461	int i;
	462
c2b3e1f7 JP	463	hfs_dbg(BNODE_REFS, "put_node(%d:%d): %d\n",
	464	node->tree->cnid, node->this,
	465	atomic_read(&node->refcnt));
4d4ef9ab	466	BUG_ON(!atomic_read(&node->refcnt));
a5e3985f	467	if (!atomic_dec_and_lock(&node->refcnt, &tree->hash_lock))
1da177e4	468	return;
1da177e4	469	for (i = 0; i < tree->pages_per_bnode; i++) {
74f9c9c2 RZ	470	if (!node->page[i])
74f9c9c2 RZ	471	continue;
1da177e4	472	mark_page_accessed(node->page[i]);
1da177e4 LT	473	}
	474
	475	if (test_bit(HFS_BNODE_DELETED, &node->flags)) {
	476	hfs_bnode_unhash(node);
	477	spin_unlock(&tree->hash_lock);
	478	hfs_bmap_free(node);
	479	hfs_bnode_free(node);
	480	return;
	481	}
	482	spin_unlock(&tree->hash_lock);
	483	}
	484	}