linux-ext4 - Re: [PATCH v3] ext4: Fix rec

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Tue, 1 Aug 2023 08:18:28 -0700
From:   "Darrick J. Wong" <djwong@...nel.org>
To:     zhangshida <starzhangzsd@...il.com>
Cc:     tytso@....edu, adilger.kernel@...ger.ca, yi.zhang@...wei.com,
        linux-ext4@...r.kernel.org, linux-kernel@...r.kernel.org,
        zhangshida@...inos.cn, stable@...nel.org,
        Andreas Dilger <adilger@...ger.ca>
Subject: Re: [PATCH v3] ext4: Fix rec_len verify error

On Tue, Aug 01, 2023 at 07:23:37PM +0800, zhangshida wrote:
> From: Shida Zhang <zhangshida@...inos.cn>
> 
> With the configuration PAGE_SIZE 64k and filesystem blocksize 64k,
> a problem occurred when more than 13 million files were directly created
> under a directory:
> 
> EXT4-fs error (device xx): ext4_dx_csum_set:492: inode #xxxx: comm xxxxx: dir seems corrupt?  Run e2fsck -D.
> EXT4-fs error (device xx): ext4_dx_csum_verify:463: inode #xxxx: comm xxxxx: dir seems corrupt?  Run e2fsck -D.
> EXT4-fs error (device xx): dx_probe:856: inode #xxxx: block 8188: comm xxxxx: Directory index failed checksum
> 
> When enough files are created, the fake_dirent->reclen will be 0xffff.
> it doesn't equal to the blocksize 65536, i.e. 0x10000.
> 
> But it is not the same condition when blocksize equals to 4k.
> when enough files are created, the fake_dirent->reclen will be 0x1000.
> it equals to the blocksize 4k, i.e. 0x1000.
> 
> The problem seems to be related to the limitation of the 16-bit field
> when the blocksize is set to 64k.
> To address this, helpers like ext4_rec_len_{from,to}_disk has already
> been introduce to complete the conversion between the encoded and the
> plain form of rec_len.
> 
> So fix this one by using the helper, and all the other
> le16_to_cpu(ext4_dir_entry{,_2}.rec_len) accesses in this file too.
> 
> Cc: stable@...nel.org
> Fixes: dbe89444042a ("ext4: Calculate and verify checksums for htree nodes")
> Suggested-by: Andreas Dilger <adilger@...ger.ca>
> Suggested-by: Darrick J. Wong <djwong@...nel.org>
> Signed-off-by: Shida Zhang <zhangshida@...inos.cn>
> ---
> v1->v2:
>  Use the existing helper to covert the rec_len, as suggested by Andreas.
> v2->v3:
>  1,Covert all the other rec_len if necessary, as suggested by Darrick.
>  2,Rephrase the commit message.
> 
>  fs/ext4/namei.c | 16 ++++++++--------
>  1 file changed, 8 insertions(+), 8 deletions(-)
> 
> diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
> index 0caf6c730ce3..8cb377b8ad86 100644
> --- a/fs/ext4/namei.c
> +++ b/fs/ext4/namei.c
> @@ -346,14 +346,14 @@ static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode,
>  
>  #ifdef PARANOID
>  	struct ext4_dir_entry *d, *top;
> +	int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);
>  
>  	d = (struct ext4_dir_entry *)bh->b_data;
>  	top = (struct ext4_dir_entry *)(bh->b_data +
> -		(EXT4_BLOCK_SIZE(inode->i_sb) -
> -		 sizeof(struct ext4_dir_entry_tail)));
> -	while (d < top && d->rec_len)
> +		(blocksize - sizeof(struct ext4_dir_entry_tail)));
> +	while (d < top && ext4_rec_len_from_disk(d->rec_len, blocksize))
>  		d = (struct ext4_dir_entry *)(((void *)d) +
> -		    le16_to_cpu(d->rec_len));
> +		    ext4_rec_len_from_disk(d->rec_len, blocksize));
>  
>  	if (d != top)
>  		return NULL;

This is sitll missing some pieces; what about this clause at line 367:

	if (t->det_reserved_zero1 ||
	    le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) ||
	    t->det_reserved_zero2 ||
	    t->det_reserved_ft != EXT4_FT_DIR_CSUM)
		return NULL;

> @@ -445,13 +445,13 @@ static struct dx_countlimit *get_dx_countlimit(struct inode *inode,
>  	struct ext4_dir_entry *dp;
>  	struct dx_root_info *root;
>  	int count_offset;
> +	int blocksize = EXT4_BLOCK_SIZE(inode->i_sb);
>  
> -	if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb))
> +	if (ext4_rec_len_from_disk(dirent->rec_len, blocksize) == blocksize)
>  		count_offset = 8;
> -	else if (le16_to_cpu(dirent->rec_len) == 12) {
> +	else if (ext4_rec_len_from_disk(dirent->rec_len, blocksize) == 12) {

Why not lift this ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ to a
local variable?  @dirent doesn't change, right?

>  		dp = (struct ext4_dir_entry *)(((void *)dirent) + 12);
> -		if (le16_to_cpu(dp->rec_len) !=
> -		    EXT4_BLOCK_SIZE(inode->i_sb) - 12)
> +		if (ext4_rec_len_from_disk(dp->rec_len, blocksize) != blocksize - 12)
>  			return NULL;
>  		root = (struct dx_root_info *)(((void *)dp + 12));
>  		if (root->reserved_zero ||

What about dx_make_map?

Here's all the opencoded access I could find:

$ git grep le16.*rec_len fs/ext4/
fs/ext4/namei.c:356:                le16_to_cpu(d->rec_len));
fs/ext4/namei.c:367:        le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) ||
fs/ext4/namei.c:449:    if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb))
fs/ext4/namei.c:451:    else if (le16_to_cpu(dirent->rec_len) == 12) {
fs/ext4/namei.c:453:            if (le16_to_cpu(dp->rec_len) !=
fs/ext4/namei.c:1338:                   map_tail->size = le16_to_cpu(de->rec_len);

--D

> -- 
> 2.27.0
>