inode.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 2,219 行 · 第 1/5 页

{
	while (p < q)
		if (*p++)
			return 0;
	return 1;
}

/**
 *	ext3_find_shared - find the indirect blocks for partial truncation.
 *	@inode:	  inode in question
 *	@depth:	  depth of the affected branch
 *	@offsets: offsets of pointers in that branch (see ext3_block_to_path)
 *	@chain:	  place to store the pointers to partial indirect blocks
 *	@top:	  place to the (detached) top of branch
 *
 *	This is a helper function used by ext3_truncate().
 *
 *	When we do truncate() we may have to clean the ends of several
 *	indirect blocks but leave the blocks themselves alive. Block is
 *	partially truncated if some data below the new i_size is refered
 *	from it (and it is on the path to the first completely truncated
 *	data block, indeed).  We have to free the top of that path along
 *	with everything to the right of the path. Since no allocation
 *	past the truncation point is possible until ext3_truncate()
 *	finishes, we may safely do the latter, but top of branch may
 *	require special attention - pageout below the truncation point
 *	might try to populate it.
 *
 *	We atomically detach the top of branch from the tree, store the
 *	block number of its root in *@top, pointers to buffer_heads of
 *	partially truncated blocks - in @chain[].bh and pointers to
 *	their last elements that should not be removed - in
 *	@chain[].p. Return value is the pointer to last filled element
 *	of @chain.
 *
 *	The work left to caller to do the actual freeing of subtrees:
 *		a) free the subtree starting from *@top
 *		b) free the subtrees whose roots are stored in
 *			(@chain[i].p+1 .. end of @chain[i].bh->b_data)
 *		c) free the subtrees growing from the inode past the @chain[0].
 *			(no partially truncated stuff there).  */

static Indirect *ext3_find_shared(struct inode *inode,
				int depth,
				int offsets[4],
				Indirect chain[4],
				__le32 *top)
{
	Indirect *partial, *p;
	int k, err;

	*top = 0;
	/* Make k index the deepest non-null offest + 1 */
	for (k = depth; k > 1 && !offsets[k-1]; k--)
		;
	partial = ext3_get_branch(inode, k, offsets, chain, &err);
	/* Writer: pointers */
	if (!partial)
		partial = chain + k-1;
	/*
	 * If the branch acquired continuation since we've looked at it -
	 * fine, it should all survive and (new) top doesn't belong to us.
	 */
	if (!partial->key && *partial->p)
		/* Writer: end */
		goto no_top;
	for (p=partial; p>chain && all_zeroes((__le32*)p->bh->b_data,p->p); p--)
		;
	/*
	 * OK, we've found the last block that must survive. The rest of our
	 * branch should be detached before unlocking. However, if that rest
	 * of branch is all ours and does not grow immediately from the inode
	 * it's easier to cheat and just decrement partial->p.
	 */
	if (p == chain + k - 1 && p > chain) {
		p->p--;
	} else {
		*top = *p->p;
		/* Nope, don't do this in ext3.  Must leave the tree intact */
#if 0
		*p->p = 0;
#endif
	}
	/* Writer: end */

	while(partial > p)
	{
		brelse(partial->bh);
		partial--;
	}
no_top:
	return partial;
}

/*
 * Zero a number of block pointers in either an inode or an indirect block.
 * If we restart the transaction we must again get write access to the
 * indirect block for further modification.
 *
 * We release `count' blocks on disk, but (last - first) may be greater
 * than `count' because there can be holes in there.
 */
static void
ext3_clear_blocks(handle_t *handle, struct inode *inode, struct buffer_head *bh,
		unsigned long block_to_free, unsigned long count,
		__le32 *first, __le32 *last)
{
	__le32 *p;
	if (try_to_extend_transaction(handle, inode)) {
		if (bh) {
			BUFFER_TRACE(bh, "call ext3_journal_dirty_metadata");
			ext3_journal_dirty_metadata(handle, bh);
		}
		ext3_mark_inode_dirty(handle, inode);
		ext3_journal_test_restart(handle, inode);
		if (bh) {
			BUFFER_TRACE(bh, "retaking write access");
			ext3_journal_get_write_access(handle, bh);
		}
	}

	/*
	 * Any buffers which are on the journal will be in memory. We find
	 * them on the hash table so journal_revoke() will run journal_forget()
	 * on them.  We've already detached each block from the file, so
	 * bforget() in journal_forget() should be safe.
	 *
	 * AKPM: turn on bforget in journal_forget()!!!
	 */
	for (p = first; p < last; p++) {
		u32 nr = le32_to_cpu(*p);
		if (nr) {
			struct buffer_head *bh;

			*p = 0;
			bh = sb_find_get_block(inode->i_sb, nr);
			ext3_forget(handle, 0, inode, bh, nr);
		}
	}

	ext3_free_blocks(handle, inode, block_to_free, count);
}

/**
 * ext3_free_data - free a list of data blocks
 * @handle:	handle for this transaction
 * @inode:	inode we are dealing with
 * @this_bh:	indirect buffer_head which contains *@first and *@last
 * @first:	array of block numbers
 * @last:	points immediately past the end of array
 *
 * We are freeing all blocks refered from that array (numbers are stored as
 * little-endian 32-bit) and updating @inode->i_blocks appropriately.
 *
 * We accumulate contiguous runs of blocks to free.  Conveniently, if these
 * blocks are contiguous then releasing them at one time will only affect one
 * or two bitmap blocks (+ group descriptor(s) and superblock) and we won't
 * actually use a lot of journal space.
 *
 * @this_bh will be %NULL if @first and @last point into the inode's direct
 * block pointers.
 */
static void ext3_free_data(handle_t *handle, struct inode *inode,
			   struct buffer_head *this_bh,
			   __le32 *first, __le32 *last)
{
	unsigned long block_to_free = 0;    /* Starting block # of a run */
	unsigned long count = 0;	    /* Number of blocks in the run */ 
	__le32 *block_to_free_p = NULL;	    /* Pointer into inode/ind
					       corresponding to
					       block_to_free */
	unsigned long nr;		    /* Current block # */
	__le32 *p;			    /* Pointer into inode/ind
					       for current block */
	int err;

	if (this_bh) {				/* For indirect block */
		BUFFER_TRACE(this_bh, "get_write_access");
		err = ext3_journal_get_write_access(handle, this_bh);
		/* Important: if we can't update the indirect pointers
		 * to the blocks, we can't free them. */
		if (err)
			return;
	}

	for (p = first; p < last; p++) {
		nr = le32_to_cpu(*p);
		if (nr) {
			/* accumulate blocks to free if they're contiguous */
			if (count == 0) {
				block_to_free = nr;
				block_to_free_p = p;
				count = 1;
			} else if (nr == block_to_free + count) {
				count++;
			} else {
				ext3_clear_blocks(handle, inode, this_bh, 
						  block_to_free,
						  count, block_to_free_p, p);
				block_to_free = nr;
				block_to_free_p = p;
				count = 1;
			}
		}
	}

	if (count > 0)
		ext3_clear_blocks(handle, inode, this_bh, block_to_free,
				  count, block_to_free_p, p);

	if (this_bh) {
		BUFFER_TRACE(this_bh, "call ext3_journal_dirty_metadata");
		ext3_journal_dirty_metadata(handle, this_bh);
	}
}

/**
 *	ext3_free_branches - free an array of branches
 *	@handle: JBD handle for this transaction
 *	@inode:	inode we are dealing with
 *	@parent_bh: the buffer_head which contains *@first and *@last
 *	@first:	array of block numbers
 *	@last:	pointer immediately past the end of array
 *	@depth:	depth of the branches to free
 *
 *	We are freeing all blocks refered from these branches (numbers are
 *	stored as little-endian 32-bit) and updating @inode->i_blocks
 *	appropriately.
 */
static void ext3_free_branches(handle_t *handle, struct inode *inode,
			       struct buffer_head *parent_bh,
			       __le32 *first, __le32 *last, int depth)
{
	unsigned long nr;
	__le32 *p;

	if (is_handle_aborted(handle))
		return;

	if (depth--) {
		struct buffer_head *bh;
		int addr_per_block = EXT3_ADDR_PER_BLOCK(inode->i_sb);
		p = last;
		while (--p >= first) {
			nr = le32_to_cpu(*p);
			if (!nr)
				continue;		/* A hole */

			/* Go read the buffer for the next level down */
			bh = sb_bread(inode->i_sb, nr);

			/*
			 * A read failure? Report error and clear slot
			 * (should be rare).
			 */
			if (!bh) {
				ext3_error(inode->i_sb, "ext3_free_branches",
					   "Read failure, inode=%ld, block=%ld",
					   inode->i_ino, nr);
				continue;
			}

			/* This zaps the entire block.  Bottom up. */
			BUFFER_TRACE(bh, "free child branches");
			ext3_free_branches(handle, inode, bh,
					   (__le32*)bh->b_data,
					   (__le32*)bh->b_data + addr_per_block,
					   depth);

			/*
			 * We've probably journalled the indirect block several
			 * times during the truncate.  But it's no longer
			 * needed and we now drop it from the transaction via
			 * journal_revoke().
			 *
			 * That's easy if it's exclusively part of this
			 * transaction.  But if it's part of the committing
			 * transaction then journal_forget() will simply
			 * brelse() it.  That means that if the underlying
			 * block is reallocated in ext3_get_block(),
			 * unmap_underlying_metadata() will find this block
			 * and will try to get rid of it.  damn, damn.
			 *
			 * If this block has already been committed to the
			 * journal, a revoke record will be written.  And
			 * revoke records must be emitted *before* clearing
			 * this block's bit in the bitmaps.
			 */
			ext3_forget(handle, 1, inode, bh, bh->b_blocknr);

			/*
			 * Everything below this this pointer has been
			 * released.  Now let this top-of-subtree go.
			 *
			 * We want the freeing of this indirect block to be
			 * atomic in the journal with the updating of the
			 * bitmap block which owns it.  So make some room in
			 * the journal.
			 *
			 * We zero the parent pointer *after* freeing its
			 * pointee in the bitmaps, so if extend_transaction()
			 * for some reason fails to put the bitmap changes and
			 * the release into the same transaction, recovery
			 * will merely complain about releasing a free block,
			 * rather than leaking blocks.
			 */
			if (is_handle_aborted(handle))
				return;
			if (try_to_extend_transaction(handle, inode)) {
				ext3_mark_inode_dirty(handle, inode);
				ext3_journal_test_restart(handle, inode);
			}

			ext3_free_blocks(handle, inode, nr, 1);

			if (parent_bh) {
				/*
				 * The block which we have just freed is
				 * pointed to by an indirect block: journal it
				 */
				BUFFER_TRACE(parent_bh, "get_write_access");
				if (!ext3_journal_get_write_access(handle,
								   parent_bh)){
					*p = 0;
					BUFFER_TRACE(parent_bh,
					"call ext3_journal_dirty_metadata");
					ext3_journal_dirty_metadata(handle, 
								    parent_bh);
				}
			}
		}
	} else {
		/* We have reached the bottom of the tree. */
		BUFFER_TRACE(parent_bh, "free data blocks");
		ext3_free_data(handle, inode, parent_bh, first, last);
	}
}

/*
 * ext3_truncate()
 *
 * We block out ext3_get_block() block instantiations across the entire
 * transaction, and VFS/VM ensures that ext3_truncate() cannot run
 * simultaneously on behalf of the same inode.
 *
 * As we work through the truncate and commmit bits of it to the journal there
 * is one core, guiding principle: the file's tree must always be consistent on
 * disk.  We must be able to restart the truncate after a crash.
 *
 * The file's tree may be transiently inconsistent in memory (although it
 * probably isn't), but whenever we close off and commit a journal transaction,
 * the contents of (the filesystem + the journal) must be consistent and
 * restartable.  It's pretty simple, really: bottom up, right to left (although
 * left-to-right works OK too).
 *
 * Note that at recovery time, journal replay occurs *before* the restart of
 * truncate against the orphan inode list.
 *
 * The committed inode has the new, desired i_size (which is the same as
 * i_disksize in this case).  After a crash, ext3_orphan_cleanup() will see
 * that this inode's truncate did not complete and it will again call
 * ext3_truncate() to have another go.  So there will be instantiated blocks
 * to the right of the truncation point in a crashed ext3 filesystem.  But
 * that's fine - as long as they are linked from the inode, the post-crash
 * ext3_truncate() run will find them and release them.
 */

void ext3_truncate(struct inode * inode)
{
	handle_t *handle;
	struct ext3_inode_info *ei = EXT3_I(inode);
	__le32 *i_data = ei->i_data;
	int addr_per_block = EXT3_ADDR_PER_BLOCK(inode->i_sb);
	struct address_space *mapping = inode->i_mapping;
	int offsets[4];
	Indirect chain[4];
	Indirect *partial;
	__le32 nr = 0;
	int n;
	long last_block;
	unsigned blocksize = inode->i_sb->s_blocksize;
	struct page *page;

	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
	    S_ISLNK(inode->i_mode)))
		return;
	if (ext3_inode_is_fast_symlink(inode))
		return;
	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
		return;

	ext3_discard_prealloc(inode);

	/*
	 * We have to lock the EOF page here, because lock_page() nests
	 * outside journal_start().
	 */
	if ((inode->i_size & (blocksize - 1)) == 0) {
		/* Block boundary? Nothing to do */
		page = NULL;
	} else {
		page = grab_cache_page(mapping,
				inode->i_size >> PAGE_CACHE_SHIFT);
		if (!page)
			return;
	}

	handle = start_transaction(inode);
	if (IS_ERR(handle)) {
		if (page) {
			clear_highpage(page);
			flush_dcache_page(page);
			unlock_page(page);
			page_cache_release(page);
		}
		return;		/* AKPM: return what? */
	}

	last_block = (inode->i_size + blocksize-1)
					>> EXT3_BLOCK_SIZE_BITS(inode->i_sb);

	if (page)
		ext3_block_truncate_page(handle, page, mapping, inode->i_size);

	n = ext3_block_to_path(inode, last_block, offsets, NULL);
	if (n == 0)
		goto out_stop;	/* error */

	/*
	 * OK.  This truncate is going to happen.  We add the inode to the
	 * orphan list, so that if this truncate spans multiple transactions,
	 * and we crash, we will resume the truncate when the filesystem
	 * recovers.  It also marks the inode dirty, to catch the new size.
	 *
	 * Implication: the file must always be in a sane, consistent
	 * truncatable state while each transaction commits.
	 */
	if (ext3_orphan_add(handle, inode))
		goto out_stop;

	/*
	 * The orphan list entry will now protect us from any crash which
	 * occurs before the trunc