2006-10-11 02:20:50 -06:00
|
|
|
/*
|
2006-10-11 02:20:53 -06:00
|
|
|
* linux/fs/ext4/dir.c
|
2006-10-11 02:20:50 -06:00
|
|
|
*
|
|
|
|
* Copyright (C) 1992, 1993, 1994, 1995
|
|
|
|
* Remy Card (card@masi.ibp.fr)
|
|
|
|
* Laboratoire MASI - Institut Blaise Pascal
|
|
|
|
* Universite Pierre et Marie Curie (Paris VI)
|
|
|
|
*
|
|
|
|
* from
|
|
|
|
*
|
|
|
|
* linux/fs/minix/dir.c
|
|
|
|
*
|
|
|
|
* Copyright (C) 1991, 1992 Linus Torvalds
|
|
|
|
*
|
2006-10-11 02:20:53 -06:00
|
|
|
* ext4 directory handling functions
|
2006-10-11 02:20:50 -06:00
|
|
|
*
|
|
|
|
* Big-endian to little-endian byte-swapping/bitmaps by
|
|
|
|
* David S. Miller (davem@caip.rutgers.edu), 1995
|
|
|
|
*
|
|
|
|
* Hash Tree Directory indexing (c) 2001 Daniel Phillips
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/fs.h>
|
2006-10-11 02:21:01 -06:00
|
|
|
#include <linux/jbd2.h>
|
2006-10-11 02:20:50 -06:00
|
|
|
#include <linux/buffer_head.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/rbtree.h>
|
2008-04-29 16:13:32 -06:00
|
|
|
#include "ext4.h"
|
2012-12-10 12:05:59 -07:00
|
|
|
#include "xattr.h"
|
2006-10-11 02:20:50 -06:00
|
|
|
|
2013-05-17 14:08:53 -06:00
|
|
|
static int ext4_dx_readdir(struct file *, struct dir_context *);
|
2006-10-11 02:20:50 -06:00
|
|
|
|
2012-03-18 20:44:40 -06:00
|
|
|
/**
|
|
|
|
* Check if the given dir-inode refers to an htree-indexed directory
|
|
|
|
* (or a directory which chould potentially get coverted to use htree
|
|
|
|
* indexing).
|
|
|
|
*
|
|
|
|
* Return 1 if it is a dx dir, 0 if not
|
|
|
|
*/
|
|
|
|
static int is_dx_dir(struct inode *inode)
|
|
|
|
{
|
|
|
|
struct super_block *sb = inode->i_sb;
|
|
|
|
|
|
|
|
if (EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
|
|
|
|
EXT4_FEATURE_COMPAT_DIR_INDEX) &&
|
|
|
|
((ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) ||
|
2013-04-19 15:53:09 -06:00
|
|
|
((inode->i_size >> sb->s_blocksize_bits) == 1) ||
|
|
|
|
ext4_has_inline_data(inode)))
|
2012-03-18 20:44:40 -06:00
|
|
|
return 1;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2010-12-19 20:07:02 -07:00
|
|
|
/*
|
|
|
|
* Return 0 if the directory entry is OK, and 1 if there is a problem
|
|
|
|
*
|
|
|
|
* Note: this is the opposite of what ext2 and ext3 historically returned...
|
2012-12-10 12:05:59 -07:00
|
|
|
*
|
|
|
|
* bh passed here can be an inode block or a dir data block, depending
|
|
|
|
* on the inode inline data flag.
|
2010-12-19 20:07:02 -07:00
|
|
|
*/
|
2010-07-27 09:54:40 -06:00
|
|
|
int __ext4_check_dir_entry(const char *function, unsigned int line,
|
2011-01-10 10:10:55 -07:00
|
|
|
struct inode *dir, struct file *filp,
|
2010-07-27 09:54:40 -06:00
|
|
|
struct ext4_dir_entry_2 *de,
|
2012-12-10 12:05:58 -07:00
|
|
|
struct buffer_head *bh, char *buf, int size,
|
2010-07-27 09:54:40 -06:00
|
|
|
unsigned int offset)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
2008-09-08 20:25:24 -06:00
|
|
|
const char *error_msg = NULL;
|
2009-02-14 21:01:36 -07:00
|
|
|
const int rlen = ext4_rec_len_from_disk(de->rec_len,
|
|
|
|
dir->i_sb->s_blocksize);
|
2006-10-11 02:20:50 -06:00
|
|
|
|
2010-12-19 20:07:02 -07:00
|
|
|
if (unlikely(rlen < EXT4_DIR_REC_LEN(1)))
|
2006-10-11 02:20:50 -06:00
|
|
|
error_msg = "rec_len is smaller than minimal";
|
2010-12-19 20:07:02 -07:00
|
|
|
else if (unlikely(rlen % 4 != 0))
|
2006-10-11 02:20:50 -06:00
|
|
|
error_msg = "rec_len % 4 != 0";
|
2010-12-19 20:07:02 -07:00
|
|
|
else if (unlikely(rlen < EXT4_DIR_REC_LEN(de->name_len)))
|
2006-10-11 02:20:50 -06:00
|
|
|
error_msg = "rec_len is too small for name_len";
|
2012-12-10 12:05:58 -07:00
|
|
|
else if (unlikely(((char *) de - buf) + rlen > size))
|
|
|
|
error_msg = "directory entry across range";
|
2010-12-19 20:07:02 -07:00
|
|
|
else if (unlikely(le32_to_cpu(de->inode) >
|
|
|
|
le32_to_cpu(EXT4_SB(dir->i_sb)->s_es->s_inodes_count)))
|
2006-10-11 02:20:50 -06:00
|
|
|
error_msg = "inode out of bounds";
|
2010-12-19 20:07:02 -07:00
|
|
|
else
|
|
|
|
return 0;
|
|
|
|
|
2011-01-10 10:10:55 -07:00
|
|
|
if (filp)
|
2012-02-20 15:53:05 -07:00
|
|
|
ext4_error_file(filp, function, line, bh->b_blocknr,
|
2011-01-10 10:10:55 -07:00
|
|
|
"bad entry in directory: %s - offset=%u(%u), "
|
|
|
|
"inode=%u, rec_len=%d, name_len=%d",
|
2012-12-10 12:05:58 -07:00
|
|
|
error_msg, (unsigned) (offset % size),
|
2011-01-10 10:10:55 -07:00
|
|
|
offset, le32_to_cpu(de->inode),
|
|
|
|
rlen, de->name_len);
|
|
|
|
else
|
2012-02-20 15:53:05 -07:00
|
|
|
ext4_error_inode(dir, function, line, bh->b_blocknr,
|
2011-01-10 10:10:55 -07:00
|
|
|
"bad entry in directory: %s - offset=%u(%u), "
|
|
|
|
"inode=%u, rec_len=%d, name_len=%d",
|
2012-12-10 12:05:58 -07:00
|
|
|
error_msg, (unsigned) (offset % size),
|
2011-01-10 10:10:55 -07:00
|
|
|
offset, le32_to_cpu(de->inode),
|
|
|
|
rlen, de->name_len);
|
|
|
|
|
2010-12-19 20:07:02 -07:00
|
|
|
return 1;
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
|
2013-05-17 14:08:53 -06:00
|
|
|
static int ext4_readdir(struct file *file, struct dir_context *ctx)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
2008-11-04 22:14:04 -07:00
|
|
|
unsigned int offset;
|
2006-10-11 02:20:50 -06:00
|
|
|
int i, stored;
|
2006-10-11 02:20:53 -06:00
|
|
|
struct ext4_dir_entry_2 *de;
|
2006-10-11 02:20:50 -06:00
|
|
|
int err;
|
2013-05-17 14:08:53 -06:00
|
|
|
struct inode *inode = file_inode(file);
|
2012-03-18 20:44:40 -06:00
|
|
|
struct super_block *sb = inode->i_sb;
|
2008-10-09 09:15:52 -06:00
|
|
|
int dir_has_error = 0;
|
2006-10-11 02:20:50 -06:00
|
|
|
|
2012-03-18 20:44:40 -06:00
|
|
|
if (is_dx_dir(inode)) {
|
2013-05-17 14:08:53 -06:00
|
|
|
err = ext4_dx_readdir(file, ctx);
|
2006-10-11 02:20:50 -06:00
|
|
|
if (err != ERR_BAD_DX_DIR) {
|
2013-05-17 14:08:53 -06:00
|
|
|
return err;
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
/*
|
|
|
|
* We don't set the inode dirty flag since it's not
|
|
|
|
* critical that it get flushed back to the disk.
|
|
|
|
*/
|
2013-05-17 14:08:53 -06:00
|
|
|
ext4_clear_inode_flag(file_inode(file),
|
2010-06-14 07:54:48 -06:00
|
|
|
EXT4_INODE_INDEX);
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
2013-04-19 15:53:09 -06:00
|
|
|
|
|
|
|
if (ext4_has_inline_data(inode)) {
|
|
|
|
int has_inline_data = 1;
|
2013-05-17 14:08:53 -06:00
|
|
|
int ret = ext4_read_inline_dir(file, ctx,
|
2013-04-19 15:53:09 -06:00
|
|
|
&has_inline_data);
|
|
|
|
if (has_inline_data)
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2006-10-11 02:20:50 -06:00
|
|
|
stored = 0;
|
2013-05-17 14:08:53 -06:00
|
|
|
offset = ctx->pos & (sb->s_blocksize - 1);
|
2006-10-11 02:20:50 -06:00
|
|
|
|
2013-05-17 14:08:53 -06:00
|
|
|
while (ctx->pos < inode->i_size) {
|
2010-05-16 18:00:00 -06:00
|
|
|
struct ext4_map_blocks map;
|
2006-10-11 02:20:50 -06:00
|
|
|
struct buffer_head *bh = NULL;
|
|
|
|
|
2013-05-17 14:08:53 -06:00
|
|
|
map.m_lblk = ctx->pos >> EXT4_BLOCK_SIZE_BITS(sb);
|
2010-05-16 18:00:00 -06:00
|
|
|
map.m_len = 1;
|
|
|
|
err = ext4_map_blocks(NULL, inode, &map, 0);
|
2006-10-11 02:20:50 -06:00
|
|
|
if (err > 0) {
|
2010-05-16 18:00:00 -06:00
|
|
|
pgoff_t index = map.m_pblk >>
|
2007-07-19 02:48:04 -06:00
|
|
|
(PAGE_CACHE_SHIFT - inode->i_blkbits);
|
2013-05-17 14:08:53 -06:00
|
|
|
if (!ra_has_index(&file->f_ra, index))
|
2007-07-19 02:48:08 -06:00
|
|
|
page_cache_sync_readahead(
|
2007-07-19 02:48:04 -06:00
|
|
|
sb->s_bdev->bd_inode->i_mapping,
|
2013-05-17 14:08:53 -06:00
|
|
|
&file->f_ra, file,
|
2007-07-19 02:48:08 -06:00
|
|
|
index, 1);
|
2013-05-17 14:08:53 -06:00
|
|
|
file->f_ra.prev_pos = (loff_t)index << PAGE_CACHE_SHIFT;
|
2010-05-16 18:00:00 -06:00
|
|
|
bh = ext4_bread(NULL, inode, map.m_lblk, 0, &err);
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We ignore I/O errors on directories so users have a chance
|
|
|
|
* of recovering data when there's a bad sector
|
|
|
|
*/
|
|
|
|
if (!bh) {
|
2008-10-09 09:15:52 -06:00
|
|
|
if (!dir_has_error) {
|
2013-05-17 14:08:53 -06:00
|
|
|
EXT4_ERROR_FILE(file, 0,
|
2011-01-10 10:10:55 -07:00
|
|
|
"directory contains a "
|
|
|
|
"hole at offset %llu",
|
2013-05-17 14:08:53 -06:00
|
|
|
(unsigned long long) ctx->pos);
|
2008-10-09 09:15:52 -06:00
|
|
|
dir_has_error = 1;
|
|
|
|
}
|
[PATCH] handle ext4 directory corruption better
I've been using Steve Grubb's purely evil "fsfuzzer" tool, at
http://people.redhat.com/sgrubb/files/fsfuzzer-0.4.tar.gz
Basically it makes a filesystem, splats some random bits over it, then
tries to mount it and do some simple filesystem actions.
At best, the filesystem catches the corruption gracefully. At worst,
things spin out of control.
As you might guess, we found a couple places in ext4 where things spin out
of control :)
First, we had a corrupted directory that was never checked for
consistency... it was corrupt, and pointed to another bad "entry" of
length 0. The for() loop looped forever, since the length of
ext4_next_entry(de) was 0, and we kept looking at the same pointer over and
over and over and over... I modeled this check and subsequent action on
what is done for other directory types in ext4_readdir...
(adding this check adds some computational expense; I am testing a followup
patch to reduce the number of times we check and re-check these directory
entries, in all cases. Thanks for the idea, Andreas).
Next we had a root directory inode which had a corrupted size, claimed to
be > 200M on a 4M filesystem. There was only really 1 block in the
directory, but because the size was so large, readdir kept coming back for
more, spewing thousands of printk's along the way.
Per Andreas' suggestion, if we're in this read error condition and we're
trying to read an offset which is greater than i_blocks worth of bytes,
stop trying, and break out of the loop.
With these two changes fsfuzz test survives quite well on ext4.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Cc: <linux-ext4@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-06 21:36:28 -07:00
|
|
|
/* corrupt size? Maybe no more blocks to read */
|
2013-05-17 14:08:53 -06:00
|
|
|
if (ctx->pos > inode->i_blocks << 9)
|
[PATCH] handle ext4 directory corruption better
I've been using Steve Grubb's purely evil "fsfuzzer" tool, at
http://people.redhat.com/sgrubb/files/fsfuzzer-0.4.tar.gz
Basically it makes a filesystem, splats some random bits over it, then
tries to mount it and do some simple filesystem actions.
At best, the filesystem catches the corruption gracefully. At worst,
things spin out of control.
As you might guess, we found a couple places in ext4 where things spin out
of control :)
First, we had a corrupted directory that was never checked for
consistency... it was corrupt, and pointed to another bad "entry" of
length 0. The for() loop looped forever, since the length of
ext4_next_entry(de) was 0, and we kept looking at the same pointer over and
over and over and over... I modeled this check and subsequent action on
what is done for other directory types in ext4_readdir...
(adding this check adds some computational expense; I am testing a followup
patch to reduce the number of times we check and re-check these directory
entries, in all cases. Thanks for the idea, Andreas).
Next we had a root directory inode which had a corrupted size, claimed to
be > 200M on a 4M filesystem. There was only really 1 block in the
directory, but because the size was so large, readdir kept coming back for
more, spewing thousands of printk's along the way.
Per Andreas' suggestion, if we're in this read error condition and we're
trying to read an offset which is greater than i_blocks worth of bytes,
stop trying, and break out of the loop.
With these two changes fsfuzz test survives quite well on ext4.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Cc: <linux-ext4@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-06 21:36:28 -07:00
|
|
|
break;
|
2013-05-17 14:08:53 -06:00
|
|
|
ctx->pos += sb->s_blocksize - offset;
|
2006-10-11 02:20:50 -06:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2012-04-29 16:41:10 -06:00
|
|
|
/* Check the checksum */
|
|
|
|
if (!buffer_verified(bh) &&
|
|
|
|
!ext4_dirent_csum_verify(inode,
|
|
|
|
(struct ext4_dir_entry *)bh->b_data)) {
|
2013-05-17 14:08:53 -06:00
|
|
|
EXT4_ERROR_FILE(file, 0, "directory fails checksum "
|
2012-04-29 16:41:10 -06:00
|
|
|
"at offset %llu",
|
2013-05-17 14:08:53 -06:00
|
|
|
(unsigned long long)ctx->pos);
|
|
|
|
ctx->pos += sb->s_blocksize - offset;
|
2013-01-28 19:23:24 -07:00
|
|
|
brelse(bh);
|
2012-04-29 16:41:10 -06:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
set_buffer_verified(bh);
|
|
|
|
|
2006-10-11 02:20:50 -06:00
|
|
|
/* If the dir block has changed since the last call to
|
|
|
|
* readdir(2), then we might be pointing to an invalid
|
|
|
|
* dirent right now. Scan from the start of the block
|
|
|
|
* to make sure. */
|
2013-05-17 14:08:53 -06:00
|
|
|
if (file->f_version != inode->i_version) {
|
2006-10-11 02:20:50 -06:00
|
|
|
for (i = 0; i < sb->s_blocksize && i < offset; ) {
|
2006-10-11 02:20:53 -06:00
|
|
|
de = (struct ext4_dir_entry_2 *)
|
2006-10-11 02:20:50 -06:00
|
|
|
(bh->b_data + i);
|
|
|
|
/* It's too expensive to do a full
|
|
|
|
* dirent test each time round this
|
|
|
|
* loop, but we do have to test at
|
|
|
|
* least that it is non-zero. A
|
|
|
|
* failure will be detected in the
|
|
|
|
* dirent test below. */
|
2009-02-14 21:01:36 -07:00
|
|
|
if (ext4_rec_len_from_disk(de->rec_len,
|
|
|
|
sb->s_blocksize) < EXT4_DIR_REC_LEN(1))
|
2006-10-11 02:20:50 -06:00
|
|
|
break;
|
2009-02-14 21:01:36 -07:00
|
|
|
i += ext4_rec_len_from_disk(de->rec_len,
|
|
|
|
sb->s_blocksize);
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
offset = i;
|
2013-05-17 14:08:53 -06:00
|
|
|
ctx->pos = (ctx->pos & ~(sb->s_blocksize - 1))
|
2006-10-11 02:20:50 -06:00
|
|
|
| offset;
|
2013-05-17 14:08:53 -06:00
|
|
|
file->f_version = inode->i_version;
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
|
2013-05-17 14:08:53 -06:00
|
|
|
while (ctx->pos < inode->i_size
|
2006-10-11 02:20:50 -06:00
|
|
|
&& offset < sb->s_blocksize) {
|
2006-10-11 02:20:53 -06:00
|
|
|
de = (struct ext4_dir_entry_2 *) (bh->b_data + offset);
|
2013-05-17 14:08:53 -06:00
|
|
|
if (ext4_check_dir_entry(inode, file, de, bh,
|
2012-12-10 12:05:58 -07:00
|
|
|
bh->b_data, bh->b_size,
|
|
|
|
offset)) {
|
2006-10-11 02:21:24 -06:00
|
|
|
/*
|
2013-05-17 14:08:53 -06:00
|
|
|
* On error, skip to the next block
|
2006-10-11 02:21:24 -06:00
|
|
|
*/
|
2013-05-17 14:08:53 -06:00
|
|
|
ctx->pos = (ctx->pos |
|
2006-10-11 02:20:50 -06:00
|
|
|
(sb->s_blocksize - 1)) + 1;
|
2013-05-17 14:08:53 -06:00
|
|
|
break;
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
2009-02-14 21:01:36 -07:00
|
|
|
offset += ext4_rec_len_from_disk(de->rec_len,
|
|
|
|
sb->s_blocksize);
|
2006-10-11 02:20:50 -06:00
|
|
|
if (le32_to_cpu(de->inode)) {
|
2013-05-17 14:08:53 -06:00
|
|
|
if (!dir_emit(ctx, de->name,
|
2006-10-11 02:20:50 -06:00
|
|
|
de->name_len,
|
|
|
|
le32_to_cpu(de->inode),
|
2013-05-17 14:08:53 -06:00
|
|
|
get_dtype(sb, de->file_type))) {
|
|
|
|
brelse(bh);
|
|
|
|
return 0;
|
|
|
|
}
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
2013-05-17 14:08:53 -06:00
|
|
|
ctx->pos += ext4_rec_len_from_disk(de->rec_len,
|
2009-02-14 21:01:36 -07:00
|
|
|
sb->s_blocksize);
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
offset = 0;
|
2008-09-08 20:25:24 -06:00
|
|
|
brelse(bh);
|
2013-05-17 14:08:53 -06:00
|
|
|
if (ctx->pos < inode->i_size) {
|
|
|
|
if (!dir_relax(inode))
|
|
|
|
return 0;
|
|
|
|
}
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
2013-05-17 14:08:53 -06:00
|
|
|
return 0;
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
|
2012-03-18 20:44:40 -06:00
|
|
|
static inline int is_32bit_api(void)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_COMPAT
|
|
|
|
return is_compat_task();
|
|
|
|
#else
|
|
|
|
return (BITS_PER_LONG == 32);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2006-10-11 02:20:50 -06:00
|
|
|
/*
|
|
|
|
* These functions convert from the major/minor hash to an f_pos
|
2012-03-18 20:44:40 -06:00
|
|
|
* value for dx directories
|
|
|
|
*
|
|
|
|
* Upper layer (for example NFS) should specify FMODE_32BITHASH or
|
|
|
|
* FMODE_64BITHASH explicitly. On the other hand, we allow ext4 to be mounted
|
|
|
|
* directly on both 32-bit and 64-bit nodes, under such case, neither
|
|
|
|
* FMODE_32BITHASH nor FMODE_64BITHASH is specified.
|
|
|
|
*/
|
|
|
|
static inline loff_t hash2pos(struct file *filp, __u32 major, __u32 minor)
|
|
|
|
{
|
|
|
|
if ((filp->f_mode & FMODE_32BITHASH) ||
|
|
|
|
(!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
|
|
|
|
return major >> 1;
|
|
|
|
else
|
|
|
|
return ((__u64)(major >> 1) << 32) | (__u64)minor;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline __u32 pos2maj_hash(struct file *filp, loff_t pos)
|
|
|
|
{
|
|
|
|
if ((filp->f_mode & FMODE_32BITHASH) ||
|
|
|
|
(!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
|
|
|
|
return (pos << 1) & 0xffffffff;
|
|
|
|
else
|
|
|
|
return ((pos >> 32) << 1) & 0xffffffff;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline __u32 pos2min_hash(struct file *filp, loff_t pos)
|
|
|
|
{
|
|
|
|
if ((filp->f_mode & FMODE_32BITHASH) ||
|
|
|
|
(!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
|
|
|
|
return 0;
|
|
|
|
else
|
|
|
|
return pos & 0xffffffff;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Return 32- or 64-bit end-of-file for dx directories
|
|
|
|
*/
|
|
|
|
static inline loff_t ext4_get_htree_eof(struct file *filp)
|
|
|
|
{
|
|
|
|
if ((filp->f_mode & FMODE_32BITHASH) ||
|
|
|
|
(!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
|
|
|
|
return EXT4_HTREE_EOF_32BIT;
|
|
|
|
else
|
|
|
|
return EXT4_HTREE_EOF_64BIT;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
2012-04-30 12:14:03 -06:00
|
|
|
* ext4_dir_llseek() calls generic_file_llseek_size to handle htree
|
|
|
|
* directories, where the "offset" is in terms of the filename hash
|
|
|
|
* value instead of the byte offset.
|
2006-10-11 02:20:50 -06:00
|
|
|
*
|
2012-04-30 12:14:03 -06:00
|
|
|
* Because we may return a 64-bit hash that is well beyond offset limits,
|
|
|
|
* we need to pass the max hash as the maximum allowable offset in
|
|
|
|
* the htree directory case.
|
|
|
|
*
|
|
|
|
* For non-htree, ext4_llseek already chooses the proper max offset.
|
2006-10-11 02:20:50 -06:00
|
|
|
*/
|
2013-03-02 15:24:05 -07:00
|
|
|
static loff_t ext4_dir_llseek(struct file *file, loff_t offset, int whence)
|
2012-03-18 20:44:40 -06:00
|
|
|
{
|
|
|
|
struct inode *inode = file->f_mapping->host;
|
|
|
|
int dx_dir = is_dx_dir(inode);
|
2012-04-30 12:14:03 -06:00
|
|
|
loff_t htree_max = ext4_get_htree_eof(file);
|
2012-03-18 20:44:40 -06:00
|
|
|
|
2012-04-30 12:14:03 -06:00
|
|
|
if (likely(dx_dir))
|
2012-12-17 16:59:39 -07:00
|
|
|
return generic_file_llseek_size(file, offset, whence,
|
2012-04-30 12:14:03 -06:00
|
|
|
htree_max, htree_max);
|
|
|
|
else
|
2012-12-17 16:59:39 -07:00
|
|
|
return ext4_llseek(file, offset, whence);
|
2012-03-18 20:44:40 -06:00
|
|
|
}
|
2006-10-11 02:20:50 -06:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This structure holds the nodes of the red-black tree used to store
|
|
|
|
* the directory entry in hash order.
|
|
|
|
*/
|
|
|
|
struct fname {
|
|
|
|
__u32 hash;
|
|
|
|
__u32 minor_hash;
|
|
|
|
struct rb_node rb_hash;
|
|
|
|
struct fname *next;
|
|
|
|
__u32 inode;
|
|
|
|
__u8 name_len;
|
|
|
|
__u8 file_type;
|
|
|
|
char name[0];
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This functoin implements a non-recursive way of freeing all of the
|
|
|
|
* nodes in the red-black tree.
|
|
|
|
*/
|
|
|
|
static void free_rb_tree_fname(struct rb_root *root)
|
|
|
|
{
|
|
|
|
struct rb_node *n = root->rb_node;
|
|
|
|
struct rb_node *parent;
|
|
|
|
struct fname *fname;
|
|
|
|
|
|
|
|
while (n) {
|
|
|
|
/* Do the node's children first */
|
2008-07-11 17:27:31 -06:00
|
|
|
if (n->rb_left) {
|
2006-10-11 02:20:50 -06:00
|
|
|
n = n->rb_left;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (n->rb_right) {
|
|
|
|
n = n->rb_right;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* The node has no children; free it, and then zero
|
|
|
|
* out parent's link to it. Finally go to the
|
|
|
|
* beginning of the loop and try to free the parent
|
|
|
|
* node.
|
|
|
|
*/
|
|
|
|
parent = rb_parent(n);
|
|
|
|
fname = rb_entry(n, struct fname, rb_hash);
|
|
|
|
while (fname) {
|
2008-09-08 20:25:24 -06:00
|
|
|
struct fname *old = fname;
|
2006-10-11 02:20:50 -06:00
|
|
|
fname = fname->next;
|
2008-09-08 20:25:24 -06:00
|
|
|
kfree(old);
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
if (!parent)
|
2010-03-04 20:25:21 -07:00
|
|
|
*root = RB_ROOT;
|
2006-10-11 02:20:50 -06:00
|
|
|
else if (parent->rb_left == n)
|
|
|
|
parent->rb_left = NULL;
|
|
|
|
else if (parent->rb_right == n)
|
|
|
|
parent->rb_right = NULL;
|
|
|
|
n = parent;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2012-03-18 20:44:40 -06:00
|
|
|
static struct dir_private_info *ext4_htree_create_dir_info(struct file *filp,
|
|
|
|
loff_t pos)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
|
|
|
struct dir_private_info *p;
|
|
|
|
|
2008-07-11 17:27:31 -06:00
|
|
|
p = kzalloc(sizeof(struct dir_private_info), GFP_KERNEL);
|
2006-10-11 02:20:50 -06:00
|
|
|
if (!p)
|
|
|
|
return NULL;
|
2012-03-18 20:44:40 -06:00
|
|
|
p->curr_hash = pos2maj_hash(filp, pos);
|
|
|
|
p->curr_minor_hash = pos2min_hash(filp, pos);
|
2006-10-11 02:20:50 -06:00
|
|
|
return p;
|
|
|
|
}
|
|
|
|
|
2006-10-11 02:20:53 -06:00
|
|
|
void ext4_htree_free_dir_info(struct dir_private_info *p)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
|
|
|
free_rb_tree_fname(&p->root);
|
|
|
|
kfree(p);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Given a directory entry, enter it into the fname rb tree.
|
|
|
|
*/
|
2006-10-11 02:20:53 -06:00
|
|
|
int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
|
2006-10-11 02:20:50 -06:00
|
|
|
__u32 minor_hash,
|
2006-10-11 02:20:53 -06:00
|
|
|
struct ext4_dir_entry_2 *dirent)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
|
|
|
struct rb_node **p, *parent = NULL;
|
2008-09-08 20:25:24 -06:00
|
|
|
struct fname *fname, *new_fn;
|
2006-10-11 02:20:50 -06:00
|
|
|
struct dir_private_info *info;
|
|
|
|
int len;
|
|
|
|
|
2010-07-27 09:56:04 -06:00
|
|
|
info = dir_file->private_data;
|
2006-10-11 02:20:50 -06:00
|
|
|
p = &info->root.rb_node;
|
|
|
|
|
|
|
|
/* Create and allocate the fname structure */
|
|
|
|
len = sizeof(struct fname) + dirent->name_len + 1;
|
|
|
|
new_fn = kzalloc(len, GFP_KERNEL);
|
|
|
|
if (!new_fn)
|
|
|
|
return -ENOMEM;
|
|
|
|
new_fn->hash = hash;
|
|
|
|
new_fn->minor_hash = minor_hash;
|
|
|
|
new_fn->inode = le32_to_cpu(dirent->inode);
|
|
|
|
new_fn->name_len = dirent->name_len;
|
|
|
|
new_fn->file_type = dirent->file_type;
|
|
|
|
memcpy(new_fn->name, dirent->name, dirent->name_len);
|
|
|
|
new_fn->name[dirent->name_len] = 0;
|
|
|
|
|
|
|
|
while (*p) {
|
|
|
|
parent = *p;
|
|
|
|
fname = rb_entry(parent, struct fname, rb_hash);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the hash and minor hash match up, then we put
|
|
|
|
* them on a linked list. This rarely happens...
|
|
|
|
*/
|
|
|
|
if ((new_fn->hash == fname->hash) &&
|
|
|
|
(new_fn->minor_hash == fname->minor_hash)) {
|
|
|
|
new_fn->next = fname->next;
|
|
|
|
fname->next = new_fn;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (new_fn->hash < fname->hash)
|
|
|
|
p = &(*p)->rb_left;
|
|
|
|
else if (new_fn->hash > fname->hash)
|
|
|
|
p = &(*p)->rb_right;
|
|
|
|
else if (new_fn->minor_hash < fname->minor_hash)
|
|
|
|
p = &(*p)->rb_left;
|
|
|
|
else /* if (new_fn->minor_hash > fname->minor_hash) */
|
|
|
|
p = &(*p)->rb_right;
|
|
|
|
}
|
|
|
|
|
|
|
|
rb_link_node(&new_fn->rb_hash, parent, p);
|
|
|
|
rb_insert_color(&new_fn->rb_hash, &info->root);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
2006-10-11 02:20:53 -06:00
|
|
|
* This is a helper function for ext4_dx_readdir. It calls filldir
|
2006-10-11 02:20:50 -06:00
|
|
|
* for all entres on the fname linked list. (Normally there is only
|
|
|
|
* one entry on the linked list, unless there are 62 bit hash collisions.)
|
|
|
|
*/
|
2013-05-17 14:08:53 -06:00
|
|
|
static int call_filldir(struct file *file, struct dir_context *ctx,
|
|
|
|
struct fname *fname)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
2013-05-17 14:08:53 -06:00
|
|
|
struct dir_private_info *info = file->private_data;
|
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
struct super_block *sb = inode->i_sb;
|
2006-10-11 02:20:50 -06:00
|
|
|
|
|
|
|
if (!fname) {
|
2012-03-19 21:41:49 -06:00
|
|
|
ext4_msg(sb, KERN_ERR, "%s:%d: inode #%lu: comm %s: "
|
|
|
|
"called with null fname?!?", __func__, __LINE__,
|
|
|
|
inode->i_ino, current->comm);
|
2006-10-11 02:20:50 -06:00
|
|
|
return 0;
|
|
|
|
}
|
2013-05-17 14:08:53 -06:00
|
|
|
ctx->pos = hash2pos(file, fname->hash, fname->minor_hash);
|
2006-10-11 02:20:50 -06:00
|
|
|
while (fname) {
|
2013-05-17 14:08:53 -06:00
|
|
|
if (!dir_emit(ctx, fname->name,
|
|
|
|
fname->name_len,
|
2006-10-11 02:20:50 -06:00
|
|
|
fname->inode,
|
2013-05-17 14:08:53 -06:00
|
|
|
get_dtype(sb, fname->file_type))) {
|
2008-08-19 19:57:43 -06:00
|
|
|
info->extra_fname = fname;
|
2013-05-17 14:08:53 -06:00
|
|
|
return 1;
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
fname = fname->next;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-05-17 14:08:53 -06:00
|
|
|
static int ext4_dx_readdir(struct file *file, struct dir_context *ctx)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
2013-05-17 14:08:53 -06:00
|
|
|
struct dir_private_info *info = file->private_data;
|
|
|
|
struct inode *inode = file_inode(file);
|
2006-10-11 02:20:50 -06:00
|
|
|
struct fname *fname;
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
if (!info) {
|
2013-05-17 14:08:53 -06:00
|
|
|
info = ext4_htree_create_dir_info(file, ctx->pos);
|
2006-10-11 02:20:50 -06:00
|
|
|
if (!info)
|
|
|
|
return -ENOMEM;
|
2013-05-17 14:08:53 -06:00
|
|
|
file->private_data = info;
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
|
2013-05-17 14:08:53 -06:00
|
|
|
if (ctx->pos == ext4_get_htree_eof(file))
|
2006-10-11 02:20:50 -06:00
|
|
|
return 0; /* EOF */
|
|
|
|
|
|
|
|
/* Some one has messed with f_pos; reset the world */
|
2013-05-17 14:08:53 -06:00
|
|
|
if (info->last_pos != ctx->pos) {
|
2006-10-11 02:20:50 -06:00
|
|
|
free_rb_tree_fname(&info->root);
|
|
|
|
info->curr_node = NULL;
|
|
|
|
info->extra_fname = NULL;
|
2013-05-17 14:08:53 -06:00
|
|
|
info->curr_hash = pos2maj_hash(file, ctx->pos);
|
|
|
|
info->curr_minor_hash = pos2min_hash(file, ctx->pos);
|
2006-10-11 02:20:50 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If there are any leftover names on the hash collision
|
|
|
|
* chain, return them first.
|
|
|
|
*/
|
2008-08-19 19:57:43 -06:00
|
|
|
if (info->extra_fname) {
|
2013-05-17 14:08:53 -06:00
|
|
|
if (call_filldir(file, ctx, info->extra_fname))
|
2008-08-19 19:57:43 -06:00
|
|
|
goto finished;
|
|
|
|
info->extra_fname = NULL;
|
2008-10-25 09:39:08 -06:00
|
|
|
goto next_node;
|
2008-08-19 19:57:43 -06:00
|
|
|
} else if (!info->curr_node)
|
2006-10-11 02:20:50 -06:00
|
|
|
info->curr_node = rb_first(&info->root);
|
|
|
|
|
|
|
|
while (1) {
|
|
|
|
/*
|
|
|
|
* Fill the rbtree if we have no more entries,
|
|
|
|
* or the inode has changed since we last read in the
|
|
|
|
* cached entries.
|
|
|
|
*/
|
|
|
|
if ((!info->curr_node) ||
|
2013-05-17 14:08:53 -06:00
|
|
|
(file->f_version != inode->i_version)) {
|
2006-10-11 02:20:50 -06:00
|
|
|
info->curr_node = NULL;
|
|
|
|
free_rb_tree_fname(&info->root);
|
2013-05-17 14:08:53 -06:00
|
|
|
file->f_version = inode->i_version;
|
|
|
|
ret = ext4_htree_fill_tree(file, info->curr_hash,
|
2006-10-11 02:20:50 -06:00
|
|
|
info->curr_minor_hash,
|
|
|
|
&info->next_hash);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
if (ret == 0) {
|
2013-05-17 14:08:53 -06:00
|
|
|
ctx->pos = ext4_get_htree_eof(file);
|
2006-10-11 02:20:50 -06:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
info->curr_node = rb_first(&info->root);
|
|
|
|
}
|
|
|
|
|
|
|
|
fname = rb_entry(info->curr_node, struct fname, rb_hash);
|
|
|
|
info->curr_hash = fname->hash;
|
|
|
|
info->curr_minor_hash = fname->minor_hash;
|
2013-05-17 14:08:53 -06:00
|
|
|
if (call_filldir(file, ctx, fname))
|
2006-10-11 02:20:50 -06:00
|
|
|
break;
|
2008-10-25 09:39:08 -06:00
|
|
|
next_node:
|
2006-10-11 02:20:50 -06:00
|
|
|
info->curr_node = rb_next(info->curr_node);
|
2008-10-25 09:39:08 -06:00
|
|
|
if (info->curr_node) {
|
|
|
|
fname = rb_entry(info->curr_node, struct fname,
|
|
|
|
rb_hash);
|
|
|
|
info->curr_hash = fname->hash;
|
|
|
|
info->curr_minor_hash = fname->minor_hash;
|
|
|
|
} else {
|
2006-10-11 02:20:50 -06:00
|
|
|
if (info->next_hash == ~0) {
|
2013-05-17 14:08:53 -06:00
|
|
|
ctx->pos = ext4_get_htree_eof(file);
|
2006-10-11 02:20:50 -06:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
info->curr_hash = info->next_hash;
|
|
|
|
info->curr_minor_hash = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
finished:
|
2013-05-17 14:08:53 -06:00
|
|
|
info->last_pos = ctx->pos;
|
2006-10-11 02:20:50 -06:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-09-08 20:25:24 -06:00
|
|
|
static int ext4_release_dir(struct inode *inode, struct file *filp)
|
2006-10-11 02:20:50 -06:00
|
|
|
{
|
2006-10-11 02:21:24 -06:00
|
|
|
if (filp->private_data)
|
2006-10-11 02:20:53 -06:00
|
|
|
ext4_htree_free_dir_info(filp->private_data);
|
2006-10-11 02:20:50 -06:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
2012-03-18 20:44:40 -06:00
|
|
|
|
|
|
|
const struct file_operations ext4_dir_operations = {
|
|
|
|
.llseek = ext4_dir_llseek,
|
|
|
|
.read = generic_read_dir,
|
2013-05-17 14:08:53 -06:00
|
|
|
.iterate = ext4_readdir,
|
2012-03-18 20:44:40 -06:00
|
|
|
.unlocked_ioctl = ext4_ioctl,
|
|
|
|
#ifdef CONFIG_COMPAT
|
|
|
|
.compat_ioctl = ext4_compat_ioctl,
|
|
|
|
#endif
|
|
|
|
.fsync = ext4_sync_file,
|
|
|
|
.release = ext4_release_dir,
|
|
|
|
};
|