[linux-block.git] / fs / hfsplus / bnode.c

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

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

#include "hfsplus_fs.h"
#include "hfsplus_raw.h"

/* Copy a specified range of bytes from the raw data of a node */
void hfs_bnode_read(struct hfs_bnode *node, void *buf, int off, int len)
{
	struct page **pagep;
	int l;

	off += node->page_offset;
	pagep = node->page + (off >> PAGE_CACHE_SHIFT);
	off &= ~PAGE_CACHE_MASK;

	l = min(len, (int)PAGE_CACHE_SIZE - off);
	memcpy(buf, kmap(*pagep) + off, l);
	kunmap(*pagep);

	while ((len -= l) != 0) {
		buf += l;
		l = min(len, (int)PAGE_CACHE_SIZE);
		memcpy(buf, kmap(*++pagep), l);
		kunmap(*pagep);
	}
}

u16 hfs_bnode_read_u16(struct hfs_bnode *node, int off)
{
	__be16 data;
	/* TODO: 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;
	/* TODO: 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_u16(node, off) + 2;
	else
		key_len = tree->max_key_len + 2;

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

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

	off += node->page_offset;
	pagep = node->page + (off >> PAGE_CACHE_SHIFT);
	off &= ~PAGE_CACHE_MASK;

	l = min(len, (int)PAGE_CACHE_SIZE - off);
	memcpy(kmap(*pagep) + off, buf, l);
	set_page_dirty(*pagep);
	kunmap(*pagep);

	while ((len -= l) != 0) {
		buf += l;
		l = min(len, (int)PAGE_CACHE_SIZE);
		memcpy(kmap(*++pagep), buf, l);
		set_page_dirty(*pagep);
		kunmap(*pagep);
	}
}

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

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

	off += node->page_offset;
	pagep = node->page + (off >> PAGE_CACHE_SHIFT);
	off &= ~PAGE_CACHE_MASK;

	l = min(len, (int)PAGE_CACHE_SIZE - off);
	memset(kmap(*pagep) + off, 0, l);
	set_page_dirty(*pagep);
	kunmap(*pagep);

	while ((len -= l) != 0) {
		l = min(len, (int)PAGE_CACHE_SIZE);
		memset(kmap(*++pagep), 0, l);
		set_page_dirty(*pagep);
		kunmap(*pagep);
	}
}

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;
	int l;

	dprint(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 + (src >> PAGE_CACHE_SHIFT);
	src &= ~PAGE_CACHE_MASK;
	dst_page = dst_node->page + (dst >> PAGE_CACHE_SHIFT);
	dst &= ~PAGE_CACHE_MASK;

	if (src == dst) {
		l = min(len, (int)PAGE_CACHE_SIZE - src);
		memcpy(kmap(*dst_page) + src, kmap(*src_page) + src, l);
		kunmap(*src_page);
		set_page_dirty(*dst_page);
		kunmap(*dst_page);

		while ((len -= l) != 0) {
			l = min(len, (int)PAGE_CACHE_SIZE);
			memcpy(kmap(*++dst_page), kmap(*++src_page), l);
			kunmap(*src_page);
			set_page_dirty(*dst_page);
			kunmap(*dst_page);
		}
	} else {
		void *src_ptr, *dst_ptr;

		do {
			src_ptr = kmap(*src_page) + src;
			dst_ptr = kmap(*dst_page) + dst;
			if (PAGE_CACHE_SIZE - src < PAGE_CACHE_SIZE - dst) {
				l = PAGE_CACHE_SIZE - src;
				src = 0;
				dst += l;
			} else {
				l = PAGE_CACHE_SIZE - dst;
				src += l;
				dst = 0;
			}
			l = min(len, l);
			memcpy(dst_ptr, src_ptr, l);
			kunmap(*src_page);
			set_page_dirty(*dst_page);
			kunmap(*dst_page);
			if (!dst)
				dst_page++;
			else
				src_page++;
		} while ((len -= l));
	}
}

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

	dprint(DBG_BNODE_MOD, "movebytes: %u,%u,%u\n", dst, src, len);
	if (!len)
		return;
	src += node->page_offset;
	dst += node->page_offset;
	if (dst > src) {
		src += len - 1;
		src_page = node->page + (src >> PAGE_CACHE_SHIFT);
		src = (src & ~PAGE_CACHE_MASK) + 1;
		dst += len - 1;
		dst_page = node->page + (dst >> PAGE_CACHE_SHIFT);
		dst = (dst & ~PAGE_CACHE_MASK) + 1;

		if (src == dst) {
			while (src < len) {
				memmove(kmap(*dst_page), kmap(*src_page), src);
				kunmap(*src_page);
				set_page_dirty(*dst_page);
				kunmap(*dst_page);
				len -= src;
				src = PAGE_CACHE_SIZE;
				src_page--;
				dst_page--;
			}
			src -= len;
			memmove(kmap(*dst_page) + src,
				kmap(*src_page) + src, len);
			kunmap(*src_page);
			set_page_dirty(*dst_page);
			kunmap(*dst_page);
		} else {
			void *src_ptr, *dst_ptr;

			do {
				src_ptr = kmap(*src_page) + src;
				dst_ptr = kmap(*dst_page) + dst;
				if (src < dst) {
					l = src;
					src = PAGE_CACHE_SIZE;
					dst -= l;
				} else {
					l = dst;
					src -= l;
					dst = PAGE_CACHE_SIZE;
				}
				l = min(len, l);
				memmove(dst_ptr - l, src_ptr - l, l);
				kunmap(*src_page);
				set_page_dirty(*dst_page);
				kunmap(*dst_page);
				if (dst == PAGE_CACHE_SIZE)
					dst_page--;
				else
					src_page--;
			} while ((len -= l));
		}
	} else {
		src_page = node->page + (src >> PAGE_CACHE_SHIFT);
		src &= ~PAGE_CACHE_MASK;
		dst_page = node->page + (dst >> PAGE_CACHE_SHIFT);
		dst &= ~PAGE_CACHE_MASK;

		if (src == dst) {
			l = min(len, (int)PAGE_CACHE_SIZE - src);
			memmove(kmap(*dst_page) + src,
				kmap(*src_page) + src, l);
			kunmap(*src_page);
			set_page_dirty(*dst_page);
			kunmap(*dst_page);

			while ((len -= l) != 0) {
				l = min(len, (int)PAGE_CACHE_SIZE);
				memmove(kmap(*++dst_page),
					kmap(*++src_page), l);
				kunmap(*src_page);
				set_page_dirty(*dst_page);
				kunmap(*dst_page);
			}
		} else {
			void *src_ptr, *dst_ptr;

			do {
				src_ptr = kmap(*src_page) + src;
				dst_ptr = kmap(*dst_page) + dst;
				if (PAGE_CACHE_SIZE - src <
						PAGE_CACHE_SIZE - dst) {
					l = PAGE_CACHE_SIZE - src;
					src = 0;
					dst += l;
				} else {
					l = PAGE_CACHE_SIZE - dst;
					src += l;
					dst = 0;
				}
				l = min(len, l);
				memmove(dst_ptr, src_ptr, l);
				kunmap(*src_page);
				set_page_dirty(*dst_page);
				kunmap(*dst_page);
				if (!dst)
					dst_page++;
				else
					src_page++;
			} while ((len -= l));
		}
	}
}

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

	dprint(DBG_BNODE_MOD, "bnode: %d\n", node->this);
	hfs_bnode_read(node, &desc, 0, sizeof(desc));
	dprint(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);
		dprint(DBG_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_u16(node, key_off) + 2;
			else
				tmp = node->tree->max_key_len + 2;
			dprint(DBG_BNODE_MOD, " (%d", tmp);
			hfs_bnode_read(node, &cnid, key_off + tmp, 4);
			dprint(DBG_BNODE_MOD, ",%d)", be32_to_cpu(cnid));
		} else if (i && node->type == HFS_NODE_LEAF) {
			int tmp;

			tmp = hfs_bnode_read_u16(node, key_off);
			dprint(DBG_BNODE_MOD, " (%d)", tmp);
		}
	}
	dprint(DBG_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) {
		dprint(DBG_BNODE_MOD, "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) {
		printk(KERN_ERR "hfs: 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) {
		printk(KERN_ERR "hfs: 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);
	dprint(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_shift;
	block = off >> PAGE_CACHE_SHIFT;
	node->page_offset = off & ~PAGE_CACHE_MASK;
	for (i = 0; i < tree->pages_per_bnode; block++, i++) {
		page = read_mapping_page(mapping, block, NULL);
		if (IS_ERR(page))
			goto fail;
		if (PageError(page)) {
			page_cache_release(page);
			goto fail;
		}
		page_cache_release(page);
		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;

	dprint(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_u16(node, off) + 2;
		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)
{
#if 0
	int i;

	for (i = 0; i < node->tree->pages_per_bnode; i++)
		if (node->page[i])
			page_cache_release(node->page[i]);
#endif
	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) {
		printk(KERN_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_CACHE_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_CACHE_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);
		dprint(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;

		dprint(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/hfsplus/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/string.h>
	12	#include <linux/slab.h>
	13	#include <linux/pagemap.h>
	14	#include <linux/fs.h>
	15	#include <linux/swap.h>
1da177e4 LT	16
	17	#include "hfsplus_fs.h"
	18	#include "hfsplus_raw.h"
	19
1da177e4 LT	20	/* Copy a specified range of bytes from the raw data of a node */
	21	void hfs_bnode_read(struct hfs_bnode node, void buf, int off, int len)
	22	{
	23	struct page **pagep;
	24	int l;
	25
	26	off += node->page_offset;
	27	pagep = node->page + (off >> PAGE_CACHE_SHIFT);
	28	off &= ~PAGE_CACHE_MASK;
	29
	30	l = min(len, (int)PAGE_CACHE_SIZE - off);
	31	memcpy(buf, kmap(*pagep) + off, l);
	32	kunmap(*pagep);
	33
	34	while ((len -= l) != 0) {
	35	buf += l;
	36	l = min(len, (int)PAGE_CACHE_SIZE);
	37	memcpy(buf, kmap(*++pagep), l);
	38	kunmap(*pagep);
	39	}
	40	}
	41
	42	u16 hfs_bnode_read_u16(struct hfs_bnode *node, int off)
	43	{
	44	__be16 data;
21f2296a	45	/* TODO: optimize later... */
1da177e4 LT	46	hfs_bnode_read(node, &data, off, 2);
	47	return be16_to_cpu(data);
	48	}
	49
	50	u8 hfs_bnode_read_u8(struct hfs_bnode *node, int off)
	51	{
	52	u8 data;
21f2296a	53	/* TODO: optimize later... */
1da177e4 LT	54	hfs_bnode_read(node, &data, off, 1);
	55	return data;
	56	}
	57
	58	void hfs_bnode_read_key(struct hfs_bnode node, void key, int off)
	59	{
	60	struct hfs_btree *tree;
	61	int key_len;
	62
	63	tree = node->tree;
	64	if (node->type == HFS_NODE_LEAF \|\|
	65	tree->attributes & HFS_TREE_VARIDXKEYS)
	66	key_len = hfs_bnode_read_u16(node, off) + 2;
	67	else
	68	key_len = tree->max_key_len + 2;
	69
	70	hfs_bnode_read(node, key, off, key_len);
	71	}
	72
	73	void hfs_bnode_write(struct hfs_bnode node, void buf, int off, int len)
	74	{
	75	struct page **pagep;
	76	int l;
	77
	78	off += node->page_offset;
	79	pagep = node->page + (off >> PAGE_CACHE_SHIFT);
	80	off &= ~PAGE_CACHE_MASK;
	81
	82	l = min(len, (int)PAGE_CACHE_SIZE - off);
	83	memcpy(kmap(*pagep) + off, buf, l);
	84	set_page_dirty(*pagep);
	85	kunmap(*pagep);
	86
	87	while ((len -= l) != 0) {
	88	buf += l;
	89	l = min(len, (int)PAGE_CACHE_SIZE);
	90	memcpy(kmap(*++pagep), buf, l);
	91	set_page_dirty(*pagep);
	92	kunmap(*pagep);
	93	}
	94	}
	95
	96	void hfs_bnode_write_u16(struct hfs_bnode *node, int off, u16 data)
	97	{
	98	__be16 v = cpu_to_be16(data);
21f2296a	99	/* TODO: optimize later... */
1da177e4 LT	100	hfs_bnode_write(node, &v, off, 2);
	101	}
	102
	103	void hfs_bnode_clear(struct hfs_bnode *node, int off, int len)
	104	{
	105	struct page **pagep;
	106	int l;
	107
	108	off += node->page_offset;
	109	pagep = node->page + (off >> PAGE_CACHE_SHIFT);
	110	off &= ~PAGE_CACHE_MASK;
	111
	112	l = min(len, (int)PAGE_CACHE_SIZE - off);
	113	memset(kmap(*pagep) + off, 0, l);
	114	set_page_dirty(*pagep);
	115	kunmap(*pagep);
	116
	117	while ((len -= l) != 0) {
	118	l = min(len, (int)PAGE_CACHE_SIZE);
	119	memset(kmap(*++pagep), 0, l);
	120	set_page_dirty(*pagep);
	121	kunmap(*pagep);
	122	}
	123	}
	124
	125	void hfs_bnode_copy(struct hfs_bnode *dst_node, int dst,
	126	struct hfs_bnode *src_node, int src, int len)
	127	{
	128	struct hfs_btree *tree;
	129	struct page src_page, dst_page;
	130	int l;
	131
	132	dprint(DBG_BNODE_MOD, "copybytes: %u,%u,%u\n", dst, src, len);
	133	if (!len)
	134	return;
	135	tree = src_node->tree;
	136	src += src_node->page_offset;
	137	dst += dst_node->page_offset;
	138	src_page = src_node->page + (src >> PAGE_CACHE_SHIFT);
	139	src &= ~PAGE_CACHE_MASK;
	140	dst_page = dst_node->page + (dst >> PAGE_CACHE_SHIFT);
	141	dst &= ~PAGE_CACHE_MASK;
	142
	143	if (src == dst) {
	144	l = min(len, (int)PAGE_CACHE_SIZE - src);
	145	memcpy(kmap(dst_page) + src, kmap(src_page) + src, l);
	146	kunmap(*src_page);
	147	set_page_dirty(*dst_page);
	148	kunmap(*dst_page);
	149
	150	while ((len -= l) != 0) {
	151	l = min(len, (int)PAGE_CACHE_SIZE);
	152	memcpy(kmap(++dst_page), kmap(++src_page), l);
	153	kunmap(*src_page);
	154	set_page_dirty(*dst_page);
	155	kunmap(*dst_page);
	156	}
	157	} else {
	158	void src_ptr, dst_ptr;
	159
	160	do {
	161	src_ptr = kmap(*src_page) + src;
	162	dst_ptr = kmap(*dst_page) + dst;
	163	if (PAGE_CACHE_SIZE - src < PAGE_CACHE_SIZE - dst) {
164	l = PAGE_CACHE_SIZE - src;
165	src = 0;
166	dst += l;
167	} else {
168	l = PAGE_CACHE_SIZE - dst;
169	src += l;
170	dst = 0;
171	}
172	l = min(len, l);
173	memcpy(dst_ptr, src_ptr, l);
174	kunmap(*src_page);
175	set_page_dirty(*dst_page);
176	kunmap(*dst_page);
177	if (!dst)
178	dst_page++;
179	else
180	src_page++;
181	} while ((len -= l));
182	}
183	}
184
185	void hfs_bnode_move(struct hfs_bnode *node, int dst, int src, int len)
186	{
187	struct page src_page, dst_page;
188	int l;
189
190	dprint(DBG_BNODE_MOD, "movebytes: %u,%u,%u\n", dst, src, len);
191	if (!len)
192	return;
193	src += node->page_offset;
194	dst += node->page_offset;
195	if (dst > src) {
196	src += len - 1;
197	src_page = node->page + (src >> PAGE_CACHE_SHIFT);
198	src = (src & ~PAGE_CACHE_MASK) + 1;
199	dst += len - 1;
200	dst_page = node->page + (dst >> PAGE_CACHE_SHIFT);
201	dst = (dst & ~PAGE_CACHE_MASK) + 1;
202
203	if (src == dst) {
204	while (src < len) {
205	memmove(kmap(dst_page), kmap(src_page), src);
206	kunmap(*src_page);
207	set_page_dirty(*dst_page);
208	kunmap(*dst_page);
209	len -= src;
210	src = PAGE_CACHE_SIZE;
211	src_page--;
212	dst_page--;
213	}
214	src -= len;
2753cc28 AS	215	memmove(kmap(*dst_page) + src,
2753cc28 AS	216	kmap(*src_page) + src, len);
1da177e4 LT	217	kunmap(*src_page);
	218	set_page_dirty(*dst_page);
	219	kunmap(*dst_page);
	220	} else {
	221	void src_ptr, dst_ptr;
	222
	223	do {
	224	src_ptr = kmap(*src_page) + src;
	225	dst_ptr = kmap(*dst_page) + dst;
	226	if (src < dst) {
	227	l = src;
	228	src = PAGE_CACHE_SIZE;
	229	dst -= l;
	230	} else {
	231	l = dst;
	232	src -= l;
	233	dst = PAGE_CACHE_SIZE;
	234	}
	235	l = min(len, l);
	236	memmove(dst_ptr - l, src_ptr - l, l);
	237	kunmap(*src_page);
	238	set_page_dirty(*dst_page);
	239	kunmap(*dst_page);
	240	if (dst == PAGE_CACHE_SIZE)
	241	dst_page--;
	242	else
	243	src_page--;
	244	} while ((len -= l));
	245	}
	246	} else {
	247	src_page = node->page + (src >> PAGE_CACHE_SHIFT);
	248	src &= ~PAGE_CACHE_MASK;
	249	dst_page = node->page + (dst >> PAGE_CACHE_SHIFT);
	250	dst &= ~PAGE_CACHE_MASK;
	251
	252	if (src == dst) {
	253	l = min(len, (int)PAGE_CACHE_SIZE - src);
2753cc28 AS	254	memmove(kmap(*dst_page) + src,
2753cc28 AS	255	kmap(*src_page) + src, l);
1da177e4 LT	256	kunmap(*src_page);
	257	set_page_dirty(*dst_page);
	258	kunmap(*dst_page);
	259
	260	while ((len -= l) != 0) {
	261	l = min(len, (int)PAGE_CACHE_SIZE);
2753cc28 AS	262	memmove(kmap(*++dst_page),
2753cc28 AS	263	kmap(*++src_page), l);
1da177e4 LT	264	kunmap(*src_page);
	265	set_page_dirty(*dst_page);
	266	kunmap(*dst_page);
	267	}
	268	} else {
	269	void src_ptr, dst_ptr;
	270
	271	do {
	272	src_ptr = kmap(*src_page) + src;
	273	dst_ptr = kmap(*dst_page) + dst;
2753cc28 AS	274	if (PAGE_CACHE_SIZE - src <
2753cc28 AS	275	PAGE_CACHE_SIZE - dst) {
1da177e4 LT	276	l = PAGE_CACHE_SIZE - src;
	277	src = 0;
	278	dst += l;
	279	} else {
	280	l = PAGE_CACHE_SIZE - dst;
	281	src += l;
	282	dst = 0;
	283	}
	284	l = min(len, l);
	285	memmove(dst_ptr, src_ptr, l);
	286	kunmap(*src_page);
	287	set_page_dirty(*dst_page);
	288	kunmap(*dst_page);
	289	if (!dst)
	290	dst_page++;
	291	else
	292	src_page++;
	293	} while ((len -= l));
	294	}
	295	}
	296	}
	297
	298	void hfs_bnode_dump(struct hfs_bnode *node)
	299	{
	300	struct hfs_bnode_desc desc;
	301	__be32 cnid;
	302	int i, off, key_off;
	303
	304	dprint(DBG_BNODE_MOD, "bnode: %d\n", node->this);
	305	hfs_bnode_read(node, &desc, 0, sizeof(desc));
	306	dprint(DBG_BNODE_MOD, "%d, %d, %d, %d, %d\n",
	307	be32_to_cpu(desc.next), be32_to_cpu(desc.prev),
	308	desc.type, desc.height, be16_to_cpu(desc.num_recs));
	309
	310	off = node->tree->node_size - 2;
	311	for (i = be16_to_cpu(desc.num_recs); i >= 0; off -= 2, i--) {
	312	key_off = hfs_bnode_read_u16(node, off);
	313	dprint(DBG_BNODE_MOD, " %d", key_off);
	314	if (i && node->type == HFS_NODE_INDEX) {
	315	int tmp;
	316
	317	if (node->tree->attributes & HFS_TREE_VARIDXKEYS)
	318	tmp = hfs_bnode_read_u16(node, key_off) + 2;
	319	else
	320	tmp = node->tree->max_key_len + 2;
	321	dprint(DBG_BNODE_MOD, " (%d", tmp);
	322	hfs_bnode_read(node, &cnid, key_off + tmp, 4);
	323	dprint(DBG_BNODE_MOD, ",%d)", be32_to_cpu(cnid));
	324	} else if (i && node->type == HFS_NODE_LEAF) {
	325	int tmp;
	326
	327	tmp = hfs_bnode_read_u16(node, key_off);
	328	dprint(DBG_BNODE_MOD, " (%d)", tmp);
	329	}
	330	}
	331	dprint(DBG_BNODE_MOD, "\n");
	332	}
	333
	334	void hfs_bnode_unlink(struct hfs_bnode *node)
	335	{
	336	struct hfs_btree *tree;
	337	struct hfs_bnode *tmp;
	338	__be32 cnid;
	339
340	tree = node->tree;
341	if (node->prev) {
342	tmp = hfs_bnode_find(tree, node->prev);
343	if (IS_ERR(tmp))
344	return;
345	tmp->next = node->next;
346	cnid = cpu_to_be32(tmp->next);
2753cc28 AS	347	hfs_bnode_write(tmp, &cnid,
2753cc28 AS	348	offsetof(struct hfs_bnode_desc, next), 4);
1da177e4 LT	349	hfs_bnode_put(tmp);
	350	} else if (node->type == HFS_NODE_LEAF)
	351	tree->leaf_head = node->next;
	352
	353	if (node->next) {
	354	tmp = hfs_bnode_find(tree, node->next);
	355	if (IS_ERR(tmp))
	356	return;
	357	tmp->prev = node->prev;
	358	cnid = cpu_to_be32(tmp->prev);
2753cc28 AS	359	hfs_bnode_write(tmp, &cnid,
2753cc28 AS	360	offsetof(struct hfs_bnode_desc, prev), 4);
1da177e4 LT	361	hfs_bnode_put(tmp);
	362	} else if (node->type == HFS_NODE_LEAF)
	363	tree->leaf_tail = node->prev;
	364
21f2296a	365	/* move down? */
1da177e4	366	if (!node->prev && !node->next) {
6d1bbfc4	367	dprint(DBG_BNODE_MOD, "hfs_btree_del_level\n");
1da177e4 LT	368	}
	369	if (!node->parent) {
	370	tree->root = 0;
	371	tree->depth = 0;
	372	}
	373	set_bit(HFS_BNODE_DELETED, &node->flags);
	374	}
	375
	376	static inline int hfs_bnode_hash(u32 num)
	377	{
	378	num = (num >> 16) + num;
	379	num += num >> 8;
	380	return num & (NODE_HASH_SIZE - 1);
	381	}
	382
	383	struct hfs_bnode hfs_bnode_findhash(struct hfs_btree tree, u32 cnid)
	384	{
	385	struct hfs_bnode *node;
	386
	387	if (cnid >= tree->node_count) {
2753cc28 AS	388	printk(KERN_ERR "hfs: request for non-existent node "
	389	"%d in B*Tree\n",
	390	cnid);
1da177e4 LT	391	return NULL;
	392	}
	393
	394	for (node = tree->node_hash[hfs_bnode_hash(cnid)];
	395	node; node = node->next_hash) {
	396	if (node->this == cnid) {
	397	return node;
	398	}
	399	}
	400	return NULL;
	401	}
	402
	403	static struct hfs_bnode __hfs_bnode_create(struct hfs_btree tree, u32 cnid)
	404	{
	405	struct super_block *sb;
	406	struct hfs_bnode node, node2;
	407	struct address_space *mapping;
	408	struct page *page;
	409	int size, block, i, hash;
	410	loff_t off;
	411
	412	if (cnid >= tree->node_count) {
2753cc28 AS	413	printk(KERN_ERR "hfs: request for non-existent node "
	414	"%d in B*Tree\n",
	415	cnid);
1da177e4 LT	416	return NULL;
	417	}
	418
	419	sb = tree->inode->i_sb;
	420	size = sizeof(struct hfs_bnode) + tree->pages_per_bnode *
	421	sizeof(struct page *);
f8314dc6	422	node = kzalloc(size, GFP_KERNEL);
1da177e4 LT	423	if (!node)
1da177e4 LT	424	return NULL;
1da177e4 LT	425	node->tree = tree;
	426	node->this = cnid;
	427	set_bit(HFS_BNODE_NEW, &node->flags);
	428	atomic_set(&node->refcnt, 1);
	429	dprint(DBG_BNODE_REFS, "new_node(%d:%d): 1\n",
	430	node->tree->cnid, node->this);
	431	init_waitqueue_head(&node->lock_wq);
	432	spin_lock(&tree->hash_lock);
	433	node2 = hfs_bnode_findhash(tree, cnid);
	434	if (!node2) {
	435	hash = hfs_bnode_hash(cnid);
	436	node->next_hash = tree->node_hash[hash];
	437	tree->node_hash[hash] = node;
	438	tree->node_hash_cnt++;
	439	} else {
	440	spin_unlock(&tree->hash_lock);
	441	kfree(node);
2753cc28 AS	442	wait_event(node2->lock_wq,
2753cc28 AS	443	!test_bit(HFS_BNODE_NEW, &node2->flags));
1da177e4 LT	444	return node2;
	445	}
	446	spin_unlock(&tree->hash_lock);
	447
	448	mapping = tree->inode->i_mapping;
	449	off = (loff_t)cnid << tree->node_size_shift;
	450	block = off >> PAGE_CACHE_SHIFT;
	451	node->page_offset = off & ~PAGE_CACHE_MASK;
	452	for (i = 0; i < tree->pages_per_bnode; block++, i++) {
090d2b18	453	page = read_mapping_page(mapping, block, NULL);
1da177e4 LT	454	if (IS_ERR(page))
	455	goto fail;
	456	if (PageError(page)) {
	457	page_cache_release(page);
	458	goto fail;
	459	}
1da177e4	460	page_cache_release(page);
1da177e4 LT	461	node->page[i] = page;
	462	}
	463
	464	return node;
	465	fail:
	466	set_bit(HFS_BNODE_ERROR, &node->flags);
	467	return node;
	468	}
	469
	470	void hfs_bnode_unhash(struct hfs_bnode *node)
	471	{
	472	struct hfs_bnode **p;
	473
	474	dprint(DBG_BNODE_REFS, "remove_node(%d:%d): %d\n",
	475	node->tree->cnid, node->this, atomic_read(&node->refcnt));
	476	for (p = &node->tree->node_hash[hfs_bnode_hash(node->this)];
	477	p && p != node; p = &(*p)->next_hash)
	478	;
0bf3ba53	479	BUG_ON(!*p);
1da177e4 LT	480	*p = node->next_hash;
	481	node->tree->node_hash_cnt--;
	482	}
	483
	484	/* Load a particular node out of a tree */
	485	struct hfs_bnode hfs_bnode_find(struct hfs_btree tree, u32 num)
	486	{
	487	struct hfs_bnode *node;
	488	struct hfs_bnode_desc *desc;
	489	int i, rec_off, off, next_off;
	490	int entry_size, key_size;
	491
	492	spin_lock(&tree->hash_lock);
	493	node = hfs_bnode_findhash(tree, num);
	494	if (node) {
	495	hfs_bnode_get(node);
	496	spin_unlock(&tree->hash_lock);
2753cc28 AS	497	wait_event(node->lock_wq,
2753cc28 AS	498	!test_bit(HFS_BNODE_NEW, &node->flags));
1da177e4 LT	499	if (test_bit(HFS_BNODE_ERROR, &node->flags))
	500	goto node_error;
	501	return node;
	502	}
	503	spin_unlock(&tree->hash_lock);
	504	node = __hfs_bnode_create(tree, num);
	505	if (!node)
	506	return ERR_PTR(-ENOMEM);
	507	if (test_bit(HFS_BNODE_ERROR, &node->flags))
	508	goto node_error;
	509	if (!test_bit(HFS_BNODE_NEW, &node->flags))
	510	return node;
	511
2753cc28 AS	512	desc = (struct hfs_bnode_desc *)(kmap(node->page[0]) +
2753cc28 AS	513	node->page_offset);
1da177e4 LT	514	node->prev = be32_to_cpu(desc->prev);
	515	node->next = be32_to_cpu(desc->next);
	516	node->num_recs = be16_to_cpu(desc->num_recs);
	517	node->type = desc->type;
	518	node->height = desc->height;
	519	kunmap(node->page[0]);
	520
	521	switch (node->type) {
	522	case HFS_NODE_HEADER:
	523	case HFS_NODE_MAP:
	524	if (node->height != 0)
	525	goto node_error;
	526	break;
	527	case HFS_NODE_LEAF:
	528	if (node->height != 1)
	529	goto node_error;
	530	break;
	531	case HFS_NODE_INDEX:
	532	if (node->height <= 1 \|\| node->height > tree->depth)
	533	goto node_error;
	534	break;
	535	default:
	536	goto node_error;
	537	}
	538
	539	rec_off = tree->node_size - 2;
	540	off = hfs_bnode_read_u16(node, rec_off);
	541	if (off != sizeof(struct hfs_bnode_desc))
	542	goto node_error;
	543	for (i = 1; i <= node->num_recs; off = next_off, i++) {
	544	rec_off -= 2;
	545	next_off = hfs_bnode_read_u16(node, rec_off);
	546	if (next_off <= off \|\|
	547	next_off > tree->node_size \|\|
	548	next_off & 1)
	549	goto node_error;
	550	entry_size = next_off - off;
	551	if (node->type != HFS_NODE_INDEX &&
	552	node->type != HFS_NODE_LEAF)
	553	continue;
	554	key_size = hfs_bnode_read_u16(node, off) + 2;
	555	if (key_size >= entry_size \|\| key_size & 1)
	556	goto node_error;
	557	}
	558	clear_bit(HFS_BNODE_NEW, &node->flags);
	559	wake_up(&node->lock_wq);
	560	return node;
	561
	562	node_error:
	563	set_bit(HFS_BNODE_ERROR, &node->flags);
	564	clear_bit(HFS_BNODE_NEW, &node->flags);
	565	wake_up(&node->lock_wq);
	566	hfs_bnode_put(node);
	567	return ERR_PTR(-EIO);
	568	}
	569
	570	void hfs_bnode_free(struct hfs_bnode *node)
	571	{
21f2296a AS	572	#if 0
21f2296a AS	573	int i;
1da177e4	574
21f2296a AS	575	for (i = 0; i < node->tree->pages_per_bnode; i++)
	576	if (node->page[i])
	577	page_cache_release(node->page[i]);
	578	#endif
1da177e4 LT	579	kfree(node);
	580	}
	581
	582	struct hfs_bnode hfs_bnode_create(struct hfs_btree tree, u32 num)
	583	{
	584	struct hfs_bnode *node;
	585	struct page **pagep;
	586	int i;
	587
	588	spin_lock(&tree->hash_lock);
	589	node = hfs_bnode_findhash(tree, num);
	590	spin_unlock(&tree->hash_lock);
	591	if (node) {
634725a9 RZ	592	printk(KERN_CRIT "new node %u already hashed?\n", num);
	593	WARN_ON(1);
	594	return node;
1da177e4 LT	595	}
	596	node = __hfs_bnode_create(tree, num);
	597	if (!node)
	598	return ERR_PTR(-ENOMEM);
	599	if (test_bit(HFS_BNODE_ERROR, &node->flags)) {
	600	hfs_bnode_put(node);
	601	return ERR_PTR(-EIO);
	602	}
	603
	604	pagep = node->page;
	605	memset(kmap(*pagep) + node->page_offset, 0,
	606	min((int)PAGE_CACHE_SIZE, (int)tree->node_size));
	607	set_page_dirty(*pagep);
	608	kunmap(*pagep);
	609	for (i = 1; i < tree->pages_per_bnode; i++) {
	610	memset(kmap(*++pagep), 0, PAGE_CACHE_SIZE);
	611	set_page_dirty(*pagep);
	612	kunmap(*pagep);
	613	}
	614	clear_bit(HFS_BNODE_NEW, &node->flags);
	615	wake_up(&node->lock_wq);
	616
	617	return node;
	618	}
	619
	620	void hfs_bnode_get(struct hfs_bnode *node)
	621	{
	622	if (node) {
	623	atomic_inc(&node->refcnt);
1da177e4	624	dprint(DBG_BNODE_REFS, "get_node(%d:%d): %d\n",
2753cc28 AS	625	node->tree->cnid, node->this,
2753cc28 AS	626	atomic_read(&node->refcnt));
1da177e4 LT	627	}
	628	}
	629
	630	/* Dispose of resources used by a node */
	631	void hfs_bnode_put(struct hfs_bnode *node)
	632	{
	633	if (node) {
	634	struct hfs_btree *tree = node->tree;
	635	int i;
	636
	637	dprint(DBG_BNODE_REFS, "put_node(%d:%d): %d\n",
2753cc28 AS	638	node->tree->cnid, node->this,
2753cc28 AS	639	atomic_read(&node->refcnt));
0bf3ba53	640	BUG_ON(!atomic_read(&node->refcnt));
a5e3985f	641	if (!atomic_dec_and_lock(&node->refcnt, &tree->hash_lock))
1da177e4	642	return;
1da177e4	643	for (i = 0; i < tree->pages_per_bnode; i++) {
74f9c9c2 RZ	644	if (!node->page[i])
74f9c9c2 RZ	645	continue;
1da177e4	646	mark_page_accessed(node->page[i]);
1da177e4 LT	647	}
	648
	649	if (test_bit(HFS_BNODE_DELETED, &node->flags)) {
	650	hfs_bnode_unhash(node);
	651	spin_unlock(&tree->hash_lock);
	652	hfs_bmap_free(node);
	653	hfs_bnode_free(node);
	654	return;
	655	}
	656	spin_unlock(&tree->hash_lock);
	657	}
	658	}
	659