dir.c

linux 内核源代码

	unmap_mft_record(dir_ni);
	m = NULL;
	ctx = NULL;
descend_into_child_node:
	/*
	 * Convert vcn to index into the index allocation attribute in units
	 * of PAGE_CACHE_SIZE and map the page cache page, reading it from
	 * disk if necessary.
	 */
	page = ntfs_map_page(ia_mapping, vcn <<
			dir_ni->itype.index.vcn_size_bits >> PAGE_CACHE_SHIFT);
	if (IS_ERR(page)) {
		ntfs_error(sb, "Failed to map directory index page, error %ld.",
				-PTR_ERR(page));
		err = PTR_ERR(page);
		goto err_out;
	}
	lock_page(page);
	kaddr = (u8*)page_address(page);
fast_descend_into_child_node:
	/* Get to the index allocation block. */
	ia = (INDEX_ALLOCATION*)(kaddr + ((vcn <<
			dir_ni->itype.index.vcn_size_bits) & ~PAGE_CACHE_MASK));
	/* Bounds checks. */
	if ((u8*)ia < kaddr || (u8*)ia > kaddr + PAGE_CACHE_SIZE) {
		ntfs_error(sb, "Out of bounds check failed. Corrupt directory "
				"inode 0x%lx or driver bug.", dir_ni->mft_no);
		goto unm_err_out;
	}
	/* Catch multi sector transfer fixup errors. */
	if (unlikely(!ntfs_is_indx_record(ia->magic))) {
		ntfs_error(sb, "Directory index record with vcn 0x%llx is "
				"corrupt.  Corrupt inode 0x%lx.  Run chkdsk.",
				(unsigned long long)vcn, dir_ni->mft_no);
		goto unm_err_out;
	}
	if (sle64_to_cpu(ia->index_block_vcn) != vcn) {
		ntfs_error(sb, "Actual VCN (0x%llx) of index buffer is "
				"different from expected VCN (0x%llx). "
				"Directory inode 0x%lx is corrupt or driver "
				"bug.", (unsigned long long)
				sle64_to_cpu(ia->index_block_vcn),
				(unsigned long long)vcn, dir_ni->mft_no);
		goto unm_err_out;
	}
	if (le32_to_cpu(ia->index.allocated_size) + 0x18 !=
			dir_ni->itype.index.block_size) {
		ntfs_error(sb, "Index buffer (VCN 0x%llx) of directory inode "
				"0x%lx has a size (%u) differing from the "
				"directory specified size (%u). Directory "
				"inode is corrupt or driver bug.",
				(unsigned long long)vcn, dir_ni->mft_no,
				le32_to_cpu(ia->index.allocated_size) + 0x18,
				dir_ni->itype.index.block_size);
		goto unm_err_out;
	}
	index_end = (u8*)ia + dir_ni->itype.index.block_size;
	if (index_end > kaddr + PAGE_CACHE_SIZE) {
		ntfs_error(sb, "Index buffer (VCN 0x%llx) of directory inode "
				"0x%lx crosses page boundary. Impossible! "
				"Cannot access! This is probably a bug in the "
				"driver.", (unsigned long long)vcn,
				dir_ni->mft_no);
		goto unm_err_out;
	}
	index_end = (u8*)&ia->index + le32_to_cpu(ia->index.index_length);
	if (index_end > (u8*)ia + dir_ni->itype.index.block_size) {
		ntfs_error(sb, "Size of index buffer (VCN 0x%llx) of directory "
				"inode 0x%lx exceeds maximum size.",
				(unsigned long long)vcn, dir_ni->mft_no);
		goto unm_err_out;
	}
	/* The first index entry. */
	ie = (INDEX_ENTRY*)((u8*)&ia->index +
			le32_to_cpu(ia->index.entries_offset));
	/*
	 * Iterate similar to above big loop but applied to index buffer, thus
	 * loop until we exceed valid memory (corruption case) or until we
	 * reach the last entry.
	 */
	for (;; ie = (INDEX_ENTRY*)((u8*)ie + le16_to_cpu(ie->length))) {
		/* Bounds check. */
		if ((u8*)ie < (u8*)ia || (u8*)ie +
				sizeof(INDEX_ENTRY_HEADER) > index_end ||
				(u8*)ie + le16_to_cpu(ie->key_length) >
				index_end) {
			ntfs_error(sb, "Index entry out of bounds in "
					"directory inode 0x%lx.",
					dir_ni->mft_no);
			goto unm_err_out;
		}
		/*
		 * The last entry cannot contain a name. It can however contain
		 * a pointer to a child node in the B+tree so we just break out.
		 */
		if (ie->flags & INDEX_ENTRY_END)
			break;
		/*
		 * If the current entry has a name type of POSIX, the name is
		 * case sensitive and not otherwise. This has the effect of us
		 * not being able to access any POSIX file names which collate
		 * after the non-POSIX one when they only differ in case, but
		 * anyone doing screwy stuff like that deserves to burn in
		 * hell... Doing that kind of stuff on NT4 actually causes
		 * corruption on the partition even when using SP6a and Linux
		 * is not involved at all.
		 */
		ic = ie->key.file_name.file_name_type ? IGNORE_CASE :
				CASE_SENSITIVE;
		/*
		 * If the names match perfectly, we are done and return the
		 * mft reference of the inode (i.e. the inode number together
		 * with the sequence number for consistency checking. We
		 * convert it to cpu format before returning.
		 */
		if (ntfs_are_names_equal(uname, uname_len,
				(ntfschar*)&ie->key.file_name.file_name,
				ie->key.file_name.file_name_length, ic,
				vol->upcase, vol->upcase_len)) {
found_it2:
			mref = le64_to_cpu(ie->data.dir.indexed_file);
			unlock_page(page);
			ntfs_unmap_page(page);
			return mref;
		}
		/*
		 * Not a perfect match, need to do full blown collation so we
		 * know which way in the B+tree we have to go.
		 */
		rc = ntfs_collate_names(uname, uname_len,
				(ntfschar*)&ie->key.file_name.file_name,
				ie->key.file_name.file_name_length, 1,
				IGNORE_CASE, vol->upcase, vol->upcase_len);
		/*
		 * If uname collates before the name of the current entry, there
		 * is definitely no such name in this index but we might need to
		 * descend into the B+tree so we just break out of the loop.
		 */
		if (rc == -1)
			break;
		/* The names are not equal, continue the search. */
		if (rc)
			continue;
		/*
		 * Names match with case insensitive comparison, now try the
		 * case sensitive comparison, which is required for proper
		 * collation.
		 */
		rc = ntfs_collate_names(uname, uname_len,
				(ntfschar*)&ie->key.file_name.file_name,
				ie->key.file_name.file_name_length, 1,
				CASE_SENSITIVE, vol->upcase, vol->upcase_len);
		if (rc == -1)
			break;
		if (rc)
			continue;
		/*
		 * Perfect match, this will never happen as the
		 * ntfs_are_names_equal() call will have gotten a match but we
		 * still treat it correctly.
		 */
		goto found_it2;
	}
	/*
	 * We have finished with this index buffer without success. Check for
	 * the presence of a child node.
	 */
	if (ie->flags & INDEX_ENTRY_NODE) {
		if ((ia->index.flags & NODE_MASK) == LEAF_NODE) {
			ntfs_error(sb, "Index entry with child node found in "
					"a leaf node in directory inode 0x%lx.",
					dir_ni->mft_no);
			goto unm_err_out;
		}
		/* Child node present, descend into it. */
		old_vcn = vcn;
		vcn = sle64_to_cpup((u8*)ie + le16_to_cpu(ie->length) - 8);
		if (vcn >= 0) {
			/* If vcn is in the same page cache page as old_vcn we
			 * recycle the mapped page. */
			if (old_vcn << vol->cluster_size_bits >>
					PAGE_CACHE_SHIFT == vcn <<
					vol->cluster_size_bits >>
					PAGE_CACHE_SHIFT)
				goto fast_descend_into_child_node;
			unlock_page(page);
			ntfs_unmap_page(page);
			goto descend_into_child_node;
		}
		ntfs_error(sb, "Negative child node vcn in directory inode "
				"0x%lx.", dir_ni->mft_no);
		goto unm_err_out;
	}
	/* No child node, return -ENOENT. */
	ntfs_debug("Entry not found.");
	err = -ENOENT;
unm_err_out:
	unlock_page(page);
	ntfs_unmap_page(page);
err_out:
	if (!err)
		err = -EIO;
	if (ctx)
		ntfs_attr_put_search_ctx(ctx);
	if (m)
		unmap_mft_record(dir_ni);
	return ERR_MREF(err);
dir_err_out:
	ntfs_error(sb, "Corrupt directory. Aborting lookup.");
	goto err_out;
}

#endif

/**
 * ntfs_filldir - ntfs specific filldir method
 * @vol:	current ntfs volume
 * @fpos:	position in the directory
 * @ndir:	ntfs inode of current directory
 * @ia_page:	page in which the index allocation buffer @ie is in resides
 * @ie:		current index entry
 * @name:	buffer to use for the converted name
 * @dirent:	vfs filldir callback context
 * @filldir:	vfs filldir callback
 *
 * Convert the Unicode @name to the loaded NLS and pass it to the @filldir
 * callback.
 *
 * If @ia_page is not NULL it is the locked page containing the index
 * allocation block containing the index entry @ie.
 *
 * Note, we drop (and then reacquire) the page lock on @ia_page across the
 * @filldir() call otherwise we would deadlock with NFSd when it calls ->lookup
 * since ntfs_lookup() will lock the same page.  As an optimization, we do not
 * retake the lock if we are returning a non-zero value as ntfs_readdir()
 * would need to drop the lock immediately anyway.
 */
static inline int ntfs_filldir(ntfs_volume *vol, loff_t fpos,
		ntfs_inode *ndir, struct page *ia_page, INDEX_ENTRY *ie,
		u8 *name, void *dirent, filldir_t filldir)
{
	unsigned long mref;
	int name_len, rc;
	unsigned dt_type;
	FILE_NAME_TYPE_FLAGS name_type;

	name_type = ie->key.file_name.file_name_type;
	if (name_type == FILE_NAME_DOS) {
		ntfs_debug("Skipping DOS name space entry.");
		return 0;
	}
	if (MREF_LE(ie->data.dir.indexed_file) == FILE_root) {
		ntfs_debug("Skipping root directory self reference entry.");
		return 0;
	}
	if (MREF_LE(ie->data.dir.indexed_file) < FILE_first_user &&
			!NVolShowSystemFiles(vol)) {
		ntfs_debug("Skipping system file.");
		return 0;
	}
	name_len = ntfs_ucstonls(vol, (ntfschar*)&ie->key.file_name.file_name,
			ie->key.file_name.file_name_length, &name,
			NTFS_MAX_NAME_LEN * NLS_MAX_CHARSET_SIZE + 1);
	if (name_len <= 0) {
		ntfs_warning(vol->sb, "Skipping unrepresentable inode 0x%llx.",
				(long long)MREF_LE(ie->data.dir.indexed_file));
		return 0;
	}
	if (ie->key.file_name.file_attributes &
			FILE_ATTR_DUP_FILE_NAME_INDEX_PRESENT)
		dt_type = DT_DIR;
	else
		dt_type = DT_REG;
	mref = MREF_LE(ie->data.dir.indexed_file);
	/*
	 * Drop the page lock otherwise we deadlock with NFS when it calls
	 * ->lookup since ntfs_lookup() will lock the same page.
	 */
	if (ia_page)
		unlock_page(ia_page);
	ntfs_debug("Calling filldir for %s with len %i, fpos 0x%llx, inode "
			"0x%lx, DT_%s.", name, name_len, fpos, mref,
			dt_type == DT_DIR ? "DIR" : "REG");
	rc = filldir(dirent, name, name_len, fpos, mref, dt_type);
	/* Relock the page but not if we are aborting ->readdir. */
	if (!rc && ia_page)
		lock_page(ia_page);
	return rc;
}

/*
 * We use the same basic approach as the old NTFS driver, i.e. we parse the
 * index root entries and then the index allocation entries that are marked
 * as in use in the index bitmap.
 *
 * While this will return the names in random order this doesn't matter for
 * ->readdir but OTOH results in a faster ->readdir.
 *
 * VFS calls ->readdir without BKL but with i_mutex held. This protects the VFS
 * parts (e.g. ->f_pos and ->i_size, and it also protects against directory
 * modifications).
 *
 * Locking:  - Caller must hold i_mutex on the directory.
 *	     - Each page cache page in the index allocation mapping must be
 *	       locked whilst being accessed otherwise we may find a corrupt
 *	       page due to it being under ->writepage at the moment which
 *	       applies the mst protection fixups before writing out and then
 *	       removes them again after the write is complete after which it 
 *	       unlocks the page.
 */
static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
{
	s64 ia_pos, ia_start, prev_ia_pos, bmp_pos;
	loff_t fpos, i_size;
	struct inode *bmp_vi, *vdir = filp->f_path.dentry->d_inode;
	struct super_block *sb = vdir->i_sb;
	ntfs_inode *ndir = NTFS_I(vdir);
	ntfs_volume *vol = NTFS_SB(sb);
	MFT_RECORD *m;
	INDEX_ROOT *ir = NULL;
	INDEX_ENTRY *ie;
	INDEX_ALLOCATION *ia;
	u8 *name = NULL;
	int rc, err, ir_pos, cur_bmp_pos;
	struct address_space *ia_mapping, *bmp_mapping;
	struct page *bmp_page = NULL, *ia_page = NULL;
	u8 *kaddr, *bmp, *index_end;
	ntfs_attr_search_ctx *ctx;

	fpos = filp->f_pos;
	ntfs_debug("Entering for inode 0x%lx, fpos 0x%llx.",
			vdir->i_ino, fpos);
	rc = err = 0;
	/* Are we at end of dir yet? */
	i_size = i_size_read(vdir);
	if (fpos >= i_size + vol->mft_record_size)
		goto done;
	/* Emulate . and .. for all directories. */
	if (!fpos) {
		ntfs_debug("Calling filldir for . with len 1, fpos 0x0, "
				"inode 0x%lx, DT_DIR.", vdir->i_ino);
		rc = filldir(dirent, ".", 1, fpos, vdir->i_ino, DT_DIR);
		if (rc)
			goto done;
		fpos++;
	}
	if (fpos == 1) {
		ntfs_debug("Calling filldir for .. with len 2, fpos 0x1, "
				"inode 0x%lx, DT_DIR.",
				(unsigned long)parent_ino(filp->f_path.dentry));
		rc = filldir(dirent, "..", 2, fpos,
				parent_ino(filp->f_path.dentry), DT_DIR);
		if (rc)
			goto done;
		fpos++;
	}
	m = NULL;
	ctx = NULL;
	/*
	 * Allocate a buffer to store the current name being processed
	 * converted to format determined by current NLS.
	 */
	name = kmalloc(NTFS_MAX_NAME_LEN * NLS_MAX_CHARSET_SIZE + 1, GFP_NOFS);
	if (unlikely(!name)) {
		err = -ENOMEM;
		goto err_out;
	}
	/* Are we jumping straight into the index allocation attribute? */
	if (fpos >= vol->mft_record_size)
		goto skip_index_root;
	/* Get hold of the mft record for the directory. */
	m = map_mft_record(ndir);
	if (IS_ERR(m)) {
		err = PTR_ERR(m);
		m = NULL;
		goto err_out;
	}
	ctx = ntfs_attr_get_search_ctx(ndir, m);
	if (unlikely(!ctx)) {
		err = -ENOMEM;
		goto err_out;
	}
	/* Get the offset into the index root attribute. */
	ir_pos = (s64)fpos;
	/* Find the index root attribute in the mft record. */
	err = ntfs_attr_lookup(AT_INDEX_ROOT, I30, 4, CASE_SENSITIVE, 0, NULL,
			0, ctx);
	if (unlikely(err)) {
		ntfs_error(sb, "Index root attribute missing in directory "
				"inode 0x%lx.", vdir->i_ino);
		goto err_out;
	}
	/*
	 * Copy the index root attribute value to a buffer so that we can put