balloc.c

《嵌入式系统设计与实例开发实验教材二源码》Linux内核移植与编译实验

/*
 *  linux/fs/ext3/balloc.c
 *
 * Copyright (C) 1992, 1993, 1994, 1995
 * Remy Card ([e-mail])
 * Laboratoire MASI - Institut Blaise Pascal
 * Universite Pierre et Marie Curie (Paris VI)
 *
 *  Enhanced block allocation by Stephen Tweedie ([e-mail]), 1993
 *  Big-endian to little-endian byte-swapping/bitmaps by
 *        David S. Miller ([e-mail]), 1995
 */

#include <linux/config.h>
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/jbd.h>
#include <linux/ext3_fs.h>
#include <linux/ext3_jbd.h>
#include <linux/locks.h>
#include <linux/quotaops.h>

/*
 * balloc.c contains the blocks allocation and deallocation routines
 */

/*
 * The free blocks are managed by bitmaps.  A file system contains several
 * blocks groups.  Each group contains 1 bitmap block for blocks, 1 bitmap
 * block for inodes, N blocks for the inode table and data blocks.
 *
 * The file system contains group descriptors which are located after the
 * super block.  Each descriptor contains the number of the bitmap block and
 * the free blocks count in the block.  The descriptors are loaded in memory
 * when a file system is mounted (see ext3_read_super).
 */


#define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)

struct ext3_group_desc * ext3_get_group_desc(struct super_block * sb,
					     unsigned int block_group,
					     struct buffer_head ** bh)
{
	unsigned long group_desc;
	unsigned long desc;
	struct ext3_group_desc * gdp;

	if (block_group >= sb->u.ext3_sb.s_groups_count) {
		ext3_error (sb, "ext3_get_group_desc",
			    "block_group >= groups_count - "
			    "block_group = %d, groups_count = %lu",
			    block_group, sb->u.ext3_sb.s_groups_count);

		return NULL;
	}
	
	group_desc = block_group / EXT3_DESC_PER_BLOCK(sb);
	desc = block_group % EXT3_DESC_PER_BLOCK(sb);
	if (!sb->u.ext3_sb.s_group_desc[group_desc]) {
		ext3_error (sb, "ext3_get_group_desc",
			    "Group descriptor not loaded - "
			    "block_group = %d, group_desc = %lu, desc = %lu",
			     block_group, group_desc, desc);
		return NULL;
	}
	
	gdp = (struct ext3_group_desc *) 
	      sb->u.ext3_sb.s_group_desc[group_desc]->b_data;
	if (bh)
		*bh = sb->u.ext3_sb.s_group_desc[group_desc];
	return gdp + desc;
}

/*
 * Read the bitmap for a given block_group, reading into the specified 
 * slot in the superblock's bitmap cache.
 *
 * Return >=0 on success or a -ve error code.
 */

static int read_block_bitmap (struct super_block * sb,
			       unsigned int block_group,
			       unsigned long bitmap_nr)
{
	struct ext3_group_desc * gdp;
	struct buffer_head * bh = NULL;
	int retval = -EIO;
	
	gdp = ext3_get_group_desc (sb, block_group, NULL);
	if (!gdp)
		goto error_out;
	retval = 0;
	bh = sb_bread(sb, le32_to_cpu(gdp->bg_block_bitmap));
	if (!bh) {
		ext3_error (sb, "read_block_bitmap",
			    "Cannot read block bitmap - "
			    "block_group = %d, block_bitmap = %lu",
			    block_group, (unsigned long) gdp->bg_block_bitmap);
		retval = -EIO;
	}
	/*
	 * On IO error, just leave a zero in the superblock's block pointer for
	 * this group.  The IO will be retried next time.
	 */
error_out:
	sb->u.ext3_sb.s_block_bitmap_number[bitmap_nr] = block_group;
	sb->u.ext3_sb.s_block_bitmap[bitmap_nr] = bh;
	return retval;
}

/*
 * load_block_bitmap loads the block bitmap for a blocks group
 *
 * It maintains a cache for the last bitmaps loaded.  This cache is managed
 * with a LRU algorithm.
 *
 * Notes:
 * 1/ There is one cache per mounted file system.
 * 2/ If the file system contains less than EXT3_MAX_GROUP_LOADED groups,
 *    this function reads the bitmap without maintaining a LRU cache.
 * 
 * Return the slot used to store the bitmap, or a -ve error code.
 */
static int __load_block_bitmap (struct super_block * sb,
			        unsigned int block_group)
{
	int i, j, retval = 0;
	unsigned long block_bitmap_number;
	struct buffer_head * block_bitmap;

	if (block_group >= sb->u.ext3_sb.s_groups_count)
		ext3_panic (sb, "load_block_bitmap",
			    "block_group >= groups_count - "
			    "block_group = %d, groups_count = %lu",
			    block_group, sb->u.ext3_sb.s_groups_count);

	if (sb->u.ext3_sb.s_groups_count <= EXT3_MAX_GROUP_LOADED) {
		if (sb->u.ext3_sb.s_block_bitmap[block_group]) {
			if (sb->u.ext3_sb.s_block_bitmap_number[block_group] ==
			    block_group)
				return block_group;
			ext3_error (sb, "__load_block_bitmap",
				    "block_group != block_bitmap_number");
		}
		retval = read_block_bitmap (sb, block_group, block_group);
		if (retval < 0)
			return retval;
		return block_group;
	}

	for (i = 0; i < sb->u.ext3_sb.s_loaded_block_bitmaps &&
		    sb->u.ext3_sb.s_block_bitmap_number[i] != block_group; i++)
		;
	if (i < sb->u.ext3_sb.s_loaded_block_bitmaps &&
  	    sb->u.ext3_sb.s_block_bitmap_number[i] == block_group) {
		block_bitmap_number = sb->u.ext3_sb.s_block_bitmap_number[i];
		block_bitmap = sb->u.ext3_sb.s_block_bitmap[i];
		for (j = i; j > 0; j--) {
			sb->u.ext3_sb.s_block_bitmap_number[j] =
				sb->u.ext3_sb.s_block_bitmap_number[j - 1];
			sb->u.ext3_sb.s_block_bitmap[j] =
				sb->u.ext3_sb.s_block_bitmap[j - 1];
		}
		sb->u.ext3_sb.s_block_bitmap_number[0] = block_bitmap_number;
		sb->u.ext3_sb.s_block_bitmap[0] = block_bitmap;

		/*
		 * There's still one special case here --- if block_bitmap == 0
		 * then our last attempt to read the bitmap failed and we have
		 * just ended up caching that failure.  Try again to read it.
		 */
		if (!block_bitmap)
			retval = read_block_bitmap (sb, block_group, 0);
	} else {
		if (sb->u.ext3_sb.s_loaded_block_bitmaps<EXT3_MAX_GROUP_LOADED)
			sb->u.ext3_sb.s_loaded_block_bitmaps++;
		else
			brelse (sb->u.ext3_sb.s_block_bitmap
					[EXT3_MAX_GROUP_LOADED - 1]);
		for (j = sb->u.ext3_sb.s_loaded_block_bitmaps - 1;
					j > 0;  j--) {
			sb->u.ext3_sb.s_block_bitmap_number[j] =
				sb->u.ext3_sb.s_block_bitmap_number[j - 1];
			sb->u.ext3_sb.s_block_bitmap[j] =
				sb->u.ext3_sb.s_block_bitmap[j - 1];
		}
		retval = read_block_bitmap (sb, block_group, 0);
	}
	return retval;
}

/*
 * Load the block bitmap for a given block group.  First of all do a couple
 * of fast lookups for common cases and then pass the request onto the guts
 * of the bitmap loader.
 *
 * Return the slot number of the group in the superblock bitmap cache's on
 * success, or a -ve error code.
 *
 * There is still one inconsistency here --- if the number of groups in this
 * filesystems is <= EXT3_MAX_GROUP_LOADED, then we have no way of 
 * differentiating between a group for which we have never performed a bitmap
 * IO request, and a group for which the last bitmap read request failed.
 */
static inline int load_block_bitmap (struct super_block * sb,
				     unsigned int block_group)
{
	int slot;
	
	/*
	 * Do the lookup for the slot.  First of all, check if we're asking
	 * for the same slot as last time, and did we succeed that last time?
	 */
	if (sb->u.ext3_sb.s_loaded_block_bitmaps > 0 &&
	    sb->u.ext3_sb.s_block_bitmap_number[0] == block_group &&
	    sb->u.ext3_sb.s_block_bitmap[0]) {
		return 0;
	}
	/*
	 * Or can we do a fast lookup based on a loaded group on a filesystem
	 * small enough to be mapped directly into the superblock?
	 */
	else if (sb->u.ext3_sb.s_groups_count <= EXT3_MAX_GROUP_LOADED && 
		 sb->u.ext3_sb.s_block_bitmap_number[block_group]==block_group
			&& sb->u.ext3_sb.s_block_bitmap[block_group]) {
		slot = block_group;
	}
	/*
	 * If not, then do a full lookup for this block group.
	 */
	else {
		slot = __load_block_bitmap (sb, block_group);
	}

	/*
	 * <0 means we just got an error
	 */
	if (slot < 0)
		return slot;
	
	/*
	 * If it's a valid slot, we may still have cached a previous IO error,
	 * in which case the bh in the superblock cache will be zero.
	 */
	if (!sb->u.ext3_sb.s_block_bitmap[slot])
		return -EIO;
	
	/*
	 * Must have been read in OK to get this far.
	 */
	return slot;
}

/* Free given blocks, update quota and i_blocks field */
void ext3_free_blocks (handle_t *handle, struct inode * inode,
			unsigned long block, unsigned long count)
{
	struct buffer_head *bitmap_bh;
	struct buffer_head *gd_bh;
	unsigned long block_group;
	unsigned long bit;
	unsigned long i;
	int bitmap_nr;
	unsigned long overflow;
	struct super_block * sb;
	struct ext3_group_desc * gdp;
	struct ext3_super_block * es;
	int err = 0, ret;
	int dquot_freed_blocks = 0;

	sb = inode->i_sb;
	if (!sb) {
		printk ("ext3_free_blocks: nonexistent device");
		return;
	}
	lock_super (sb);
	es = sb->u.ext3_sb.s_es;
	if (block < le32_to_cpu(es->s_first_data_block) || 
	    (block + count) > le32_to_cpu(es->s_blocks_count)) {
		ext3_error (sb, "ext3_free_blocks",
			    "Freeing blocks not in datazone - "
			    "block = %lu, count = %lu", block, count);
		goto error_return;
	}

	ext3_debug ("freeing block %lu\n", block);

do_more:
	overflow = 0;
	block_group = (block - le32_to_cpu(es->s_first_data_block)) /
		      EXT3_BLOCKS_PER_GROUP(sb);
	bit = (block - le32_to_cpu(es->s_first_data_block)) %
		      EXT3_BLOCKS_PER_GROUP(sb);
	/*
	 * Check to see if we are freeing blocks across a group
	 * boundary.
	 */
	if (bit + count > EXT3_BLOCKS_PER_GROUP(sb)) {
		overflow = bit + count - EXT3_BLOCKS_PER_GROUP(sb);
		count -= overflow;
	}
	bitmap_nr = load_block_bitmap (sb, block_group);
	if (bitmap_nr < 0)
		goto error_return;
	
	bitmap_bh = sb->u.ext3_sb.s_block_bitmap[bitmap_nr];
	gdp = ext3_get_group_desc (sb, block_group, &gd_bh);
	if (!gdp)
		goto error_return;

	if (in_range (le32_to_cpu(gdp->bg_block_bitmap), block, count) ||
	    in_range (le32_to_cpu(gdp->bg_inode_bitmap), block, count) ||
	    in_range (block, le32_to_cpu(gdp->bg_inode_table),
		      sb->u.ext3_sb.s_itb_per_group) ||
	    in_range (block + count - 1, le32_to_cpu(gdp->bg_inode_table),
		      sb->u.ext3_sb.s_itb_per_group))
		ext3_error (sb, "ext3_free_blocks",
			    "Freeing blocks in system zones - "
			    "Block = %lu, count = %lu",
			    block, count);

	/*
	 * We are about to start releasing blocks in the bitmap,
	 * so we need undo access.
	 */
	/* @@@ check errors */
	BUFFER_TRACE(bitmap_bh, "getting undo access");
	err = ext3_journal_get_undo_access(handle, bitmap_bh);
	if (err)
		goto error_return;
	
	/*
	 * We are about to modify some metadata.  Call the journal APIs
	 * to unshare ->b_data if a currently-committing transaction is
	 * using it
	 */
	BUFFER_TRACE(gd_bh, "get_write_access");
	err = ext3_journal_get_write_access(handle, gd_bh);	
	if (err)
		goto error_return;

	BUFFER_TRACE(sb->u.ext3_sb.s_sbh, "get_write_access");
	err = ext3_journal_get_write_access(handle, sb->u.ext3_sb.s_sbh);
	if (err)
		goto error_return;

	for (i = 0; i < count; i++) {
		/*
		 * An HJ special.  This is expensive...
		 */
#ifdef CONFIG_JBD_DEBUG
		{
			struct buffer_head *debug_bh;
			debug_bh = sb_get_hash_table(sb, block + i);
			if (debug_bh) {
				BUFFER_TRACE(debug_bh, "Deleted!");
				if (!bh2jh(bitmap_bh)->b_committed_data)
					BUFFER_TRACE(debug_bh,
						"No commited data in bitmap");
				BUFFER_TRACE2(debug_bh, bitmap_bh, "bitmap");
				__brelse(debug_bh);
			}
		}
#endif
		BUFFER_TRACE(bitmap_bh, "clear bit");
		if (!ext3_clear_bit (bit + i, bitmap_bh->b_data)) {
			ext3_error (sb, __FUNCTION__,
				      "bit already cleared for block %lu", 
				      block + i);
			BUFFER_TRACE(bitmap_bh, "bit already cleared");
		} else {
			dquot_freed_blocks++;
			gdp->bg_free_blocks_count =
			  cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count)+1);
			es->s_free_blocks_count =
			  cpu_to_le32(le32_to_cpu(es->s_free_blocks_count)+1);
		}
		/* @@@ This prevents newly-allocated data from being
		 * freed and then reallocated within the same
		 * transaction. 
		 * 
		 * Ideally we would want to allow that to happen, but to
		 * do so requires making journal_forget() capable of
		 * revoking the queued write of a data block, which
		 * implies blocking on the journal lock.  *forget()
		 * cannot block due to truncate races.
		 *
		 * Eventually we can fix this by making journal_forget()
		 * return a status indicating whether or not it was able
		 * to revoke the buffer.  On successful revoke, it is
		 * safe not to set the allocation bit in the committed
		 * bitmap, because we know that there is no outstanding
		 * activity on the buffer any more and so it is safe to
		 * reallocate it.  
		 */
		BUFFER_TRACE(bitmap_bh, "clear in b_committed_data");
		J_ASSERT_BH(bitmap_bh,
				bh2jh(bitmap_bh)->b_committed_data != NULL);
		ext3_set_bit(bit + i, bh2jh(bitmap_bh)->b_committed_data);
	}

	/* We dirtied the bitmap block */
	BUFFER_TRACE(bitmap_bh, "dirtied bitmap block");
	err = ext3_journal_dirty_metadata(handle, bitmap_bh);

	/* And the group descriptor block */
	BUFFER_TRACE(gd_bh, "dirtied group descriptor block");
	ret = ext3_journal_dirty_metadata(handle, gd_bh);
	if (!err) err = ret;

	/* And the superblock */
	BUFFER_TRACE(sb->u.ext3_sb.s_sbh, "dirtied superblock");
	ret = ext3_journal_dirty_metadata(handle, sb->u.ext3_sb.s_sbh);
	if (!err) err = ret;

	if (overflow && !err) {
		block += count;
		count = overflow;
		goto do_more;
	}
	sb->s_dirt = 1;
error_return:
	ext3_std_error(sb, err);
	unlock_super(sb);
	if (dquot_freed_blocks)
		DQUOT_FREE_BLOCK(inode, dquot_freed_blocks);
	return;
}

/* For ext3 allocations, we must not reuse any blocks which are
 * allocated in the bitmap buffer's "last committed data" copy.  This
 * prevents deletes from freeing up the page for reuse until we have
 * committed the delete transaction.
 *
 * If we didn't do this, then deleting something and reallocating it as
 * data would allow the old block to be overwritten before the
 * transaction committed (because we force data to disk before commit).
 * This would lead to corruption if we crashed between overwriting the
 * data and committing the delete. 
 *
 * @@@ We may want to make this allocation behaviour conditional on
 * data-writes at some point, and disable it for metadata allocations or
 * sync-data inodes.
 */
static int ext3_test_allocatable(int nr, struct buffer_head *bh)
{
	if (ext3_test_bit(nr, bh->b_data))
		return 0;
	if (!buffer_jbd(bh) || !bh2jh(bh)->b_committed_data)
		return 1;
	return !ext3_test_bit(nr, bh2jh(bh)->b_committed_data);
}

/*
 * Find an allocatable block in a bitmap.  We honour both the bitmap and
 * its last-committed copy (if that exists), and perform the "most
 * appropriate allocation" algorithm of looking for a free block near
 * the initial goal; then for a free byte somewhere in the bitmap; then
 * for any free bit in the bitmap.
 */
static int find_next_usable_block(int start,
			struct buffer_head *bh, int maxblocks)
{
	int here, next;
	char *p, *r;
	
	if (start > 0) {
		/*
		 * The goal was occupied; search forward for a free 
		 * block within the next XX blocks.
		 *
		 * end_goal is more or less random, but it has to be
		 * less than EXT3_BLOCKS_PER_GROUP. Aligning up to the
		 * next 64-bit boundary is simple..
		 */
		int end_goal = (start + 63) & ~63;
		here = ext3_find_next_zero_bit(bh->b_data, end_goal, start);
		if (here < end_goal && ext3_test_allocatable(here, bh))
			return here;
		
		ext3_debug ("Bit not found near goal\n");
		
	}
	
	here = start;
	if (here < 0)
		here = 0;
	
	/*
	 * There has been no free block found in the near vicinity of
	 * the goal: do a search forward through the block groups,
	 * searching in each group first for an entire free byte in the
	 * bitmap and then for any free bit.
	 * 
	 * Search first in the remainder of the current group 
	 */
	p = ((char *) bh->b_data) + (here >> 3);
	r = memscan(p, 0, (maxblocks - here + 7) >> 3);
	next = (r - ((char *) bh->b_data)) << 3;