[linux.git] / fs / ext4 / block_validity.c

/*
 *  linux/fs/ext4/block_validity.c
 *
 * Copyright (C) 2009
 * Theodore Ts'o ([email protected])
 *
 * Track which blocks in the filesystem are metadata blocks that
 * should never be used as data blocks by files or directories.
 */

#include <linux/time.h>
#include <linux/fs.h>
#include <linux/namei.h>
#include <linux/quotaops.h>
#include <linux/buffer_head.h>
#include <linux/swap.h>
#include <linux/pagemap.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
#include "ext4.h"

struct ext4_system_zone {
	struct rb_node	node;
	ext4_fsblk_t	start_blk;
	unsigned int	count;
};

static struct kmem_cache *ext4_system_zone_cachep;

int __init ext4_init_system_zone(void)
{
	ext4_system_zone_cachep = KMEM_CACHE(ext4_system_zone, 0);
	if (ext4_system_zone_cachep == NULL)
		return -ENOMEM;
	return 0;
}

void ext4_exit_system_zone(void)
{
	kmem_cache_destroy(ext4_system_zone_cachep);
}

static inline int can_merge(struct ext4_system_zone *entry1,
		     struct ext4_system_zone *entry2)
{
	if ((entry1->start_blk + entry1->count) == entry2->start_blk)
		return 1;
	return 0;
}

/*
 * Mark a range of blocks as belonging to the "system zone" --- that
 * is, filesystem metadata blocks which should never be used by
 * inodes.
 */
static int add_system_zone(struct ext4_sb_info *sbi,
			   ext4_fsblk_t start_blk,
			   unsigned int count)
{
	struct ext4_system_zone *new_entry = NULL, *entry;
	struct rb_node **n = &sbi->system_blks.rb_node, *node;
	struct rb_node *parent = NULL, *new_node = NULL;

	while (*n) {
		parent = *n;
		entry = rb_entry(parent, struct ext4_system_zone, node);
		if (start_blk < entry->start_blk)
			n = &(*n)->rb_left;
		else if (start_blk >= (entry->start_blk + entry->count))
			n = &(*n)->rb_right;
		else {
			if (start_blk + count > (entry->start_blk +
						 entry->count))
				entry->count = (start_blk + count -
						entry->start_blk);
			new_node = *n;
			new_entry = rb_entry(new_node, struct ext4_system_zone,
					     node);
			break;
		}
	}

	if (!new_entry) {
		new_entry = kmem_cache_alloc(ext4_system_zone_cachep,
					     GFP_KERNEL);
		if (!new_entry)
			return -ENOMEM;
		new_entry->start_blk = start_blk;
		new_entry->count = count;
		new_node = &new_entry->node;

		rb_link_node(new_node, parent, n);
		rb_insert_color(new_node, &sbi->system_blks);
	}

	/* Can we merge to the left? */
	node = rb_prev(new_node);
	if (node) {
		entry = rb_entry(node, struct ext4_system_zone, node);
		if (can_merge(entry, new_entry)) {
			new_entry->start_blk = entry->start_blk;
			new_entry->count += entry->count;
			rb_erase(node, &sbi->system_blks);
			kmem_cache_free(ext4_system_zone_cachep, entry);
		}
	}

	/* Can we merge to the right? */
	node = rb_next(new_node);
	if (node) {
		entry = rb_entry(node, struct ext4_system_zone, node);
		if (can_merge(new_entry, entry)) {
			new_entry->count += entry->count;
			rb_erase(node, &sbi->system_blks);
			kmem_cache_free(ext4_system_zone_cachep, entry);
		}
	}
	return 0;
}

static void debug_print_tree(struct ext4_sb_info *sbi)
{
	struct rb_node *node;
	struct ext4_system_zone *entry;
	int first = 1;

	printk(KERN_INFO "System zones: ");
	node = rb_first(&sbi->system_blks);
	while (node) {
		entry = rb_entry(node, struct ext4_system_zone, node);
		printk(KERN_CONT "%s%llu-%llu", first ? "" : ", ",
		       entry->start_blk, entry->start_blk + entry->count - 1);
		first = 0;
		node = rb_next(node);
	}
	printk(KERN_CONT "\n");
}

int ext4_setup_system_zone(struct super_block *sb)
{
	ext4_group_t ngroups = ext4_get_groups_count(sb);
	struct ext4_sb_info *sbi = EXT4_SB(sb);
	struct ext4_group_desc *gdp;
	ext4_group_t i;
	int flex_size = ext4_flex_bg_size(sbi);
	int ret;

	if (!test_opt(sb, BLOCK_VALIDITY)) {
		if (EXT4_SB(sb)->system_blks.rb_node)
			ext4_release_system_zone(sb);
		return 0;
	}
	if (EXT4_SB(sb)->system_blks.rb_node)
		return 0;

	for (i=0; i < ngroups; i++) {
		if (ext4_bg_has_super(sb, i) &&
		    ((i < 5) || ((i % flex_size) == 0)))
			add_system_zone(sbi, ext4_group_first_block_no(sb, i),
					ext4_bg_num_gdb(sb, i) + 1);
		gdp = ext4_get_group_desc(sb, i, NULL);
		ret = add_system_zone(sbi, ext4_block_bitmap(sb, gdp), 1);
		if (ret)
			return ret;
		ret = add_system_zone(sbi, ext4_inode_bitmap(sb, gdp), 1);
		if (ret)
			return ret;
		ret = add_system_zone(sbi, ext4_inode_table(sb, gdp),
				sbi->s_itb_per_group);
		if (ret)
			return ret;
	}

	if (test_opt(sb, DEBUG))
		debug_print_tree(EXT4_SB(sb));
	return 0;
}

/* Called when the filesystem is unmounted */
void ext4_release_system_zone(struct super_block *sb)
{
	struct ext4_system_zone	*entry, *n;

	rbtree_postorder_for_each_entry_safe(entry, n,
			&EXT4_SB(sb)->system_blks, node)
		kmem_cache_free(ext4_system_zone_cachep, entry);

	EXT4_SB(sb)->system_blks = RB_ROOT;
}

/*
 * Returns 1 if the passed-in block region (start_blk,
 * start_blk+count) is valid; 0 if some part of the block region
 * overlaps with filesystem metadata blocks.
 */
int ext4_data_block_valid(struct ext4_sb_info *sbi, ext4_fsblk_t start_blk,
			  unsigned int count)
{
	struct ext4_system_zone *entry;
	struct rb_node *n = sbi->system_blks.rb_node;

	if ((start_blk <= le32_to_cpu(sbi->s_es->s_first_data_block)) ||
	    (start_blk + count < start_blk) ||
	    (start_blk + count > ext4_blocks_count(sbi->s_es))) {
		sbi->s_es->s_last_error_block = cpu_to_le64(start_blk);
		return 0;
	}
	while (n) {
		entry = rb_entry(n, struct ext4_system_zone, node);
		if (start_blk + count - 1 < entry->start_blk)
			n = n->rb_left;
		else if (start_blk >= (entry->start_blk + entry->count))
			n = n->rb_right;
		else {
			sbi->s_es->s_last_error_block = cpu_to_le64(start_blk);
			return 0;
		}
	}
	return 1;
}

int ext4_check_blockref(const char *function, unsigned int line,
			struct inode *inode, __le32 *p, unsigned int max)
{
	struct ext4_super_block *es = EXT4_SB(inode->i_sb)->s_es;
	__le32 *bref = p;
	unsigned int blk;

	while (bref < p+max) {
		blk = le32_to_cpu(*bref++);
		if (blk &&
		    unlikely(!ext4_data_block_valid(EXT4_SB(inode->i_sb),
						    blk, 1))) {
			es->s_last_error_block = cpu_to_le64(blk);
			ext4_error_inode(inode, function, line, blk,
					 "invalid block");
			return -EFSCORRUPTED;
		}
	}
	return 0;
}
Commit	Line	Data
6fd058f7 TT	1	/*
	2	* linux/fs/ext4/block_validity.c
	3	*
	4	* Copyright (C) 2009
	5	* Theodore Ts'o ([email protected])
	6	*
	7	* Track which blocks in the filesystem are metadata blocks that
	8	* should never be used as data blocks by files or directories.
	9	*/
	10
	11	#include <linux/time.h>
	12	#include <linux/fs.h>
	13	#include <linux/namei.h>
	14	#include <linux/quotaops.h>
	15	#include <linux/buffer_head.h>
6fd058f7 TT	16	#include <linux/swap.h>
6fd058f7 TT	17	#include <linux/pagemap.h>
6fd058f7	18	#include <linux/blkdev.h>
5a0e3ad6	19	#include <linux/slab.h>
6fd058f7 TT	20	#include "ext4.h"
	21
	22	struct ext4_system_zone {
	23	struct rb_node node;
	24	ext4_fsblk_t start_blk;
	25	unsigned int count;
	26	};
	27
	28	static struct kmem_cache *ext4_system_zone_cachep;
	29
5dabfc78	30	int __init ext4_init_system_zone(void)
6fd058f7	31	{
16828088	32	ext4_system_zone_cachep = KMEM_CACHE(ext4_system_zone, 0);
6fd058f7 TT	33	if (ext4_system_zone_cachep == NULL)
	34	return -ENOMEM;
	35	return 0;
	36	}
	37
5dabfc78	38	void ext4_exit_system_zone(void)
6fd058f7 TT	39	{
	40	kmem_cache_destroy(ext4_system_zone_cachep);
	41	}
	42
	43	static inline int can_merge(struct ext4_system_zone *entry1,
	44	struct ext4_system_zone *entry2)
	45	{
	46	if ((entry1->start_blk + entry1->count) == entry2->start_blk)
	47	return 1;
	48	return 0;
	49	}
	50
	51	/*
	52	* Mark a range of blocks as belonging to the "system zone" --- that
	53	* is, filesystem metadata blocks which should never be used by
	54	* inodes.
	55	*/
	56	static int add_system_zone(struct ext4_sb_info *sbi,
	57	ext4_fsblk_t start_blk,
	58	unsigned int count)
	59	{
	60	struct ext4_system_zone new_entry = NULL, entry;
	61	struct rb_node *n = &sbi->system_blks.rb_node, node;
	62	struct rb_node parent = NULL, new_node = NULL;
	63
	64	while (*n) {
	65	parent = *n;
	66	entry = rb_entry(parent, struct ext4_system_zone, node);
	67	if (start_blk < entry->start_blk)
	68	n = &(*n)->rb_left;
	69	else if (start_blk >= (entry->start_blk + entry->count))
	70	n = &(*n)->rb_right;
	71	else {
60e6679e	72	if (start_blk + count > (entry->start_blk +
6fd058f7	73	entry->count))
60e6679e	74	entry->count = (start_blk + count -
6fd058f7 TT	75	entry->start_blk);
	76	new_node = *n;
	77	new_entry = rb_entry(new_node, struct ext4_system_zone,
	78	node);
	79	break;
	80	}
	81	}
	82
	83	if (!new_entry) {
	84	new_entry = kmem_cache_alloc(ext4_system_zone_cachep,
	85	GFP_KERNEL);
	86	if (!new_entry)
	87	return -ENOMEM;
	88	new_entry->start_blk = start_blk;
	89	new_entry->count = count;
	90	new_node = &new_entry->node;
	91
	92	rb_link_node(new_node, parent, n);
	93	rb_insert_color(new_node, &sbi->system_blks);
	94	}
	95
	96	/* Can we merge to the left? */
	97	node = rb_prev(new_node);
	98	if (node) {
	99	entry = rb_entry(node, struct ext4_system_zone, node);
	100	if (can_merge(entry, new_entry)) {
	101	new_entry->start_blk = entry->start_blk;
	102	new_entry->count += entry->count;
	103	rb_erase(node, &sbi->system_blks);
	104	kmem_cache_free(ext4_system_zone_cachep, entry);
	105	}
	106	}
	107
	108	/* Can we merge to the right? */
	109	node = rb_next(new_node);
	110	if (node) {
	111	entry = rb_entry(node, struct ext4_system_zone, node);
	112	if (can_merge(new_entry, entry)) {
	113	new_entry->count += entry->count;
	114	rb_erase(node, &sbi->system_blks);
	115	kmem_cache_free(ext4_system_zone_cachep, entry);
	116	}
	117	}
	118	return 0;
	119	}
	120
	121	static void debug_print_tree(struct ext4_sb_info *sbi)
	122	{
	123	struct rb_node *node;
	124	struct ext4_system_zone *entry;
	125	int first = 1;
	126
	127	printk(KERN_INFO "System zones: ");
	128	node = rb_first(&sbi->system_blks);
	129	while (node) {
	130	entry = rb_entry(node, struct ext4_system_zone, node);
d74f3d25	131	printk(KERN_CONT "%s%llu-%llu", first ? "" : ", ",
6fd058f7 TT	132	entry->start_blk, entry->start_blk + entry->count - 1);
	133	first = 0;
	134	node = rb_next(node);
	135	}
d74f3d25	136	printk(KERN_CONT "\n");
6fd058f7 TT	137	}
	138
	139	int ext4_setup_system_zone(struct super_block *sb)
	140	{
	141	ext4_group_t ngroups = ext4_get_groups_count(sb);
	142	struct ext4_sb_info *sbi = EXT4_SB(sb);
	143	struct ext4_group_desc *gdp;
	144	ext4_group_t i;
	145	int flex_size = ext4_flex_bg_size(sbi);
	146	int ret;
	147
	148	if (!test_opt(sb, BLOCK_VALIDITY)) {
	149	if (EXT4_SB(sb)->system_blks.rb_node)
	150	ext4_release_system_zone(sb);
	151	return 0;
	152	}
	153	if (EXT4_SB(sb)->system_blks.rb_node)
	154	return 0;
	155
	156	for (i=0; i < ngroups; i++) {
	157	if (ext4_bg_has_super(sb, i) &&
	158	((i < 5) \|\| ((i % flex_size) == 0)))
	159	add_system_zone(sbi, ext4_group_first_block_no(sb, i),
1032988c	160	ext4_bg_num_gdb(sb, i) + 1);
6fd058f7 TT	161	gdp = ext4_get_group_desc(sb, i, NULL);
	162	ret = add_system_zone(sbi, ext4_block_bitmap(sb, gdp), 1);
	163	if (ret)
	164	return ret;
	165	ret = add_system_zone(sbi, ext4_inode_bitmap(sb, gdp), 1);
	166	if (ret)
	167	return ret;
	168	ret = add_system_zone(sbi, ext4_inode_table(sb, gdp),
	169	sbi->s_itb_per_group);
	170	if (ret)
	171	return ret;
	172	}
	173
	174	if (test_opt(sb, DEBUG))
	175	debug_print_tree(EXT4_SB(sb));
	176	return 0;
	177	}
	178
	179	/* Called when the filesystem is unmounted */
	180	void ext4_release_system_zone(struct super_block *sb)
	181	{
d1866bd0 CS	182	struct ext4_system_zone entry, n;
	183
	184	rbtree_postorder_for_each_entry_safe(entry, n,
	185	&EXT4_SB(sb)->system_blks, node)
6fd058f7	186	kmem_cache_free(ext4_system_zone_cachep, entry);
d1866bd0	187
64e290ec	188	EXT4_SB(sb)->system_blks = RB_ROOT;
6fd058f7 TT	189	}
	190
	191	/*
	192	* Returns 1 if the passed-in block region (start_blk,
	193	* start_blk+count) is valid; 0 if some part of the block region
	194	* overlaps with filesystem metadata blocks.
	195	*/
	196	int ext4_data_block_valid(struct ext4_sb_info *sbi, ext4_fsblk_t start_blk,
	197	unsigned int count)
	198	{
	199	struct ext4_system_zone *entry;
	200	struct rb_node *n = sbi->system_blks.rb_node;
	201
	202	if ((start_blk <= le32_to_cpu(sbi->s_es->s_first_data_block)) \|\|
1585d8d8	203	(start_blk + count < start_blk) \|\|
1c13d5c0 TT	204	(start_blk + count > ext4_blocks_count(sbi->s_es))) {
1c13d5c0 TT	205	sbi->s_es->s_last_error_block = cpu_to_le64(start_blk);
6fd058f7	206	return 0;
1c13d5c0	207	}
6fd058f7 TT	208	while (n) {
	209	entry = rb_entry(n, struct ext4_system_zone, node);
	210	if (start_blk + count - 1 < entry->start_blk)
	211	n = n->rb_left;
	212	else if (start_blk >= (entry->start_blk + entry->count))
	213	n = n->rb_right;
1c13d5c0 TT	214	else {
1c13d5c0 TT	215	sbi->s_es->s_last_error_block = cpu_to_le64(start_blk);
6fd058f7	216	return 0;
1c13d5c0	217	}
6fd058f7 TT	218	}
	219	return 1;
	220	}
	221
1f7d1e77 TT	222	int ext4_check_blockref(const char *function, unsigned int line,
	223	struct inode inode, __le32 p, unsigned int max)
	224	{
	225	struct ext4_super_block *es = EXT4_SB(inode->i_sb)->s_es;
	226	__le32 *bref = p;
	227	unsigned int blk;
	228
	229	while (bref < p+max) {
	230	blk = le32_to_cpu(*bref++);
	231	if (blk &&
	232	unlikely(!ext4_data_block_valid(EXT4_SB(inode->i_sb),
	233	blk, 1))) {
	234	es->s_last_error_block = cpu_to_le64(blk);
	235	ext4_error_inode(inode, function, line, blk,
	236	"invalid block");
6a797d27	237	return -EFSCORRUPTED;
1f7d1e77 TT	238	}
	239	}
	240	return 0;
	241	}
dae1e52c	242