dir.c

linux 内核源代码

/**
 * dir.c - NTFS kernel directory operations. Part of the Linux-NTFS project.
 *
 * Copyright (c) 2001-2007 Anton Altaparmakov
 * Copyright (c) 2002 Richard Russon
 *
 * This program/include file is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as published
 * by the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program/include file is distributed in the hope that it will be
 * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program (in the main directory of the Linux-NTFS
 * distribution in the file COPYING); if not, write to the Free Software
 * Foundation,Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */

#include <linux/buffer_head.h>

#include "dir.h"
#include "aops.h"
#include "attrib.h"
#include "mft.h"
#include "debug.h"
#include "ntfs.h"

/**
 * The little endian Unicode string $I30 as a global constant.
 */
ntfschar I30[5] = { const_cpu_to_le16('$'), const_cpu_to_le16('I'),
		const_cpu_to_le16('3'),	const_cpu_to_le16('0'), 0 };

/**
 * ntfs_lookup_inode_by_name - find an inode in a directory given its name
 * @dir_ni:	ntfs inode of the directory in which to search for the name
 * @uname:	Unicode name for which to search in the directory
 * @uname_len:	length of the name @uname in Unicode characters
 * @res:	return the found file name if necessary (see below)
 *
 * Look for an inode with name @uname in the directory with inode @dir_ni.
 * ntfs_lookup_inode_by_name() walks the contents of the directory looking for
 * the Unicode name. If the name is found in the directory, the corresponding
 * inode number (>= 0) is returned as a mft reference in cpu format, i.e. it
 * is a 64-bit number containing the sequence number.
 *
 * On error, a negative value is returned corresponding to the error code. In
 * particular if the inode is not found -ENOENT is returned. Note that you
 * can't just check the return value for being negative, you have to check the
 * inode number for being negative which you can extract using MREC(return
 * value).
 *
 * Note, @uname_len does not include the (optional) terminating NULL character.
 *
 * Note, we look for a case sensitive match first but we also look for a case
 * insensitive match at the same time. If we find a case insensitive match, we
 * save that for the case that we don't find an exact match, where we return
 * the case insensitive match and setup @res (which we allocate!) with the mft
 * reference, the file name type, length and with a copy of the little endian
 * Unicode file name itself. If we match a file name which is in the DOS name
 * space, we only return the mft reference and file name type in @res.
 * ntfs_lookup() then uses this to find the long file name in the inode itself.
 * This is to avoid polluting the dcache with short file names. We want them to
 * work but we don't care for how quickly one can access them. This also fixes
 * the dcache aliasing issues.
 *
 * 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.
 */
MFT_REF ntfs_lookup_inode_by_name(ntfs_inode *dir_ni, const ntfschar *uname,
		const int uname_len, ntfs_name **res)
{
	ntfs_volume *vol = dir_ni->vol;
	struct super_block *sb = vol->sb;
	MFT_RECORD *m;
	INDEX_ROOT *ir;
	INDEX_ENTRY *ie;
	INDEX_ALLOCATION *ia;
	u8 *index_end;
	u64 mref;
	ntfs_attr_search_ctx *ctx;
	int err, rc;
	VCN vcn, old_vcn;
	struct address_space *ia_mapping;
	struct page *page;
	u8 *kaddr;
	ntfs_name *name = NULL;

	BUG_ON(!S_ISDIR(VFS_I(dir_ni)->i_mode));
	BUG_ON(NInoAttr(dir_ni));
	/* Get hold of the mft record for the directory. */
	m = map_mft_record(dir_ni);
	if (IS_ERR(m)) {
		ntfs_error(sb, "map_mft_record() failed with error code %ld.",
				-PTR_ERR(m));
		return ERR_MREF(PTR_ERR(m));
	}
	ctx = ntfs_attr_get_search_ctx(dir_ni, m);
	if (unlikely(!ctx)) {
		err = -ENOMEM;
		goto err_out;
	}
	/* 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)) {
		if (err == -ENOENT) {
			ntfs_error(sb, "Index root attribute missing in "
					"directory inode 0x%lx.",
					dir_ni->mft_no);
			err = -EIO;
		}
		goto err_out;
	}
	/* Get to the index root value (it's been verified in read_inode). */
	ir = (INDEX_ROOT*)((u8*)ctx->attr +
			le16_to_cpu(ctx->attr->data.resident.value_offset));
	index_end = (u8*)&ir->index + le32_to_cpu(ir->index.index_length);
	/* The first index entry. */
	ie = (INDEX_ENTRY*)((u8*)&ir->index +
			le32_to_cpu(ir->index.entries_offset));
	/*
	 * 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 checks. */
		if ((u8*)ie < (u8*)ctx->mrec || (u8*)ie +
				sizeof(INDEX_ENTRY_HEADER) > index_end ||
				(u8*)ie + le16_to_cpu(ie->key_length) >
				index_end)
			goto dir_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;
		/*
		 * We perform a case sensitive comparison and if that matches
		 * 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,
				CASE_SENSITIVE, vol->upcase, vol->upcase_len)) {
found_it:
			/*
			 * We have a perfect match, so we don't need to care
			 * about having matched imperfectly before, so we can
			 * free name and set *res to NULL.
			 * However, if the perfect match is a short file name,
			 * we need to signal this through *res, so that
			 * ntfs_lookup() can fix dcache aliasing issues.
			 * As an optimization we just reuse an existing
			 * allocation of *res.
			 */
			if (ie->key.file_name.file_name_type == FILE_NAME_DOS) {
				if (!name) {
					name = kmalloc(sizeof(ntfs_name),
							GFP_NOFS);
					if (!name) {
						err = -ENOMEM;
						goto err_out;
					}
				}
				name->mref = le64_to_cpu(
						ie->data.dir.indexed_file);
				name->type = FILE_NAME_DOS;
				name->len = 0;
				*res = name;
			} else {
				kfree(name);
				*res = NULL;
			}
			mref = le64_to_cpu(ie->data.dir.indexed_file);
			ntfs_attr_put_search_ctx(ctx);
			unmap_mft_record(dir_ni);
			return mref;
		}
		/*
		 * For a case insensitive mount, we also perform a case
		 * insensitive comparison (provided the file name is not in the
		 * POSIX namespace). If the comparison matches, and the name is
		 * in the WIN32 namespace, we cache the filename in *res so
		 * that the caller, ntfs_lookup(), can work on it. If the
		 * comparison matches, and the name is in the DOS namespace, we
		 * only cache the mft reference and the file name type (we set
		 * the name length to zero for simplicity).
		 */
		if (!NVolCaseSensitive(vol) &&
				ie->key.file_name.file_name_type &&
				ntfs_are_names_equal(uname, uname_len,
				(ntfschar*)&ie->key.file_name.file_name,
				ie->key.file_name.file_name_length,
				IGNORE_CASE, vol->upcase, vol->upcase_len)) {
			int name_size = sizeof(ntfs_name);
			u8 type = ie->key.file_name.file_name_type;
			u8 len = ie->key.file_name.file_name_length;

			/* Only one case insensitive matching name allowed. */
			if (name) {
				ntfs_error(sb, "Found already allocated name "
						"in phase 1. Please run chkdsk "
						"and if that doesn't find any "
						"errors please report you saw "
						"this message to "
						"linux-ntfs-dev@lists."
						"sourceforge.net.");
				goto dir_err_out;
			}

			if (type != FILE_NAME_DOS)
				name_size += len * sizeof(ntfschar);
			name = kmalloc(name_size, GFP_NOFS);
			if (!name) {
				err = -ENOMEM;
				goto err_out;
			}
			name->mref = le64_to_cpu(ie->data.dir.indexed_file);
			name->type = type;
			if (type != FILE_NAME_DOS) {
				name->len = len;
				memcpy(name->name, ie->key.file_name.file_name,
						len * sizeof(ntfschar));
			} else
				name->len = 0;
			*res = name;
		}
		/*
		 * 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_it;
	}
	/*
	 * We have finished with this index without success. Check for the
	 * presence of a child node and if not present return -ENOENT, unless
	 * we have got a matching name cached in name in which case return the
	 * mft reference associated with it.
	 */
	if (!(ie->flags & INDEX_ENTRY_NODE)) {
		if (name) {
			ntfs_attr_put_search_ctx(ctx);
			unmap_mft_record(dir_ni);
			return name->mref;
		}
		ntfs_debug("Entry not found.");
		err = -ENOENT;
		goto err_out;
	} /* Child node present, descend into it. */
	/* Consistency check: Verify that an index allocation exists. */
	if (!NInoIndexAllocPresent(dir_ni)) {
		ntfs_error(sb, "No index allocation attribute but index entry "
				"requires one. Directory inode 0x%lx is "
				"corrupt or driver bug.", dir_ni->mft_no);
		goto err_out;
	}
	/* Get the starting vcn of the index_block holding the child node. */
	vcn = sle64_to_cpup((sle64*)((u8*)ie + le16_to_cpu(ie->length) - 8));
	ia_mapping = VFS_I(dir_ni)->i_mapping;
	/*
	 * We are done with the index root and the mft record. Release them,
	 * otherwise we deadlock with ntfs_map_page().
	 */
	ntfs_attr_put_search_ctx(ctx);
	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 ||