linux-ext4 - Re: [PATCH v4 04/10] ext4: Implement ci comparison using unicode

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]
Message-ID: <Ynycm9QGS7MIU4io@sol.localdomain>
Date:   Wed, 11 May 2022 22:35:23 -0700
From:   Eric Biggers <ebiggers@...nel.org>
To:     Gabriel Krisman Bertazi <krisman@...labora.com>
Cc:     tytso@....edu, adilger.kernel@...ger.ca, jaegeuk@...nel.org,
        linux-ext4@...r.kernel.org, linux-f2fs-devel@...ts.sourceforge.net,
        kernel@...labora.com
Subject: Re: [PATCH v4 04/10] ext4: Implement ci comparison using unicode_name

On Wed, May 11, 2022 at 03:31:40PM -0400, Gabriel Krisman Bertazi wrote:
> By using a new type here, we can hide most of the caching casefold logic
> from ext4.  The condition in ext4_match is now quite redundant, but this
> is addressed in the next patch.
> 
> This doesn't use ext4_filename to keep it generic, since the function
> will be moved to libfs to be shared with f2fs.
> 
> Signed-off-by: Gabriel Krisman Bertazi <krisman@...labora.com>
> 
> --
> Changes since v1:
>   - Instead of (ab)using fscrypt_name, create a new type (ebiggers).
> ---
>  fs/ext4/namei.c    | 32 +++++++++++++++-----------------
>  include/linux/fs.h |  5 +++++
>  2 files changed, 20 insertions(+), 17 deletions(-)
> 
> diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
> index 84fdb23f09b8..5296ced2e43e 100644
> --- a/fs/ext4/namei.c
> +++ b/fs/ext4/namei.c
> @@ -1321,20 +1321,19 @@ static void dx_insert_block(struct dx_frame *frame, u32 hash, ext4_lblk_t block)
>  /**
>   * ext4_match_ci() - Match (case-insensitive) a name with a dirent.
>   * @parent: Inode of the parent of the dentry.
> - * @name: name under lookup.
> + * @uname: name under lookup.
>   * @de_name: Dirent name.
>   * @de_name_len: dirent name length.
> - * @quick: whether @name is already casefolded.
>   *
>   * Test whether a case-insensitive directory entry matches the filename
> - * being searched.  If quick is set, the @name being looked up is
> - * already in the casefolded form.
> + * being searched.
>   *
>   * Return: > 0 if the directory entry matches, 0 if it doesn't match, or
>   * < 0 on error.
>   */
> -static int ext4_match_ci(const struct inode *parent, const struct qstr *name,
> -			 u8 *de_name, size_t de_name_len, bool quick)
> +static int ext4_match_ci(const struct inode *parent,
> +			 const struct unicode_name *uname,
> +			 u8 *de_name, size_t de_name_len)
>  {
>  	const struct super_block *sb = parent->i_sb;
>  	const struct unicode_map *um = sb->s_encoding;
> @@ -1357,10 +1356,10 @@ static int ext4_match_ci(const struct inode *parent, const struct qstr *name,
>  		entry.len = decrypted_name.len;
>  	}
>  
> -	if (quick)
> -		ret = utf8_strncasecmp_folded(um, name, &entry);
> +	if (uname->folded_name->name)
> +		ret = utf8_strncasecmp_folded(um, uname->folded_name, &entry);
>  	else
> -		ret = utf8_strncasecmp(um, name, &entry);
> +		ret = utf8_strncasecmp(um, uname->usr_name, &entry);
>  
>  	if (!ret)
>  		match = true;
> @@ -1370,8 +1369,8 @@ static int ext4_match_ci(const struct inode *parent, const struct qstr *name,
>  		 * the names have invalid characters.
>  		 */
>  		ret = 0;
> -		match = ((name->len == entry.len) &&
> -			 !memcmp(name->name, entry.name, entry.len));
> +		match = ((uname->usr_name->len == entry.len) &&
> +			 !memcmp(uname->usr_name->name, entry.name, entry.len));
>  	}
>  
>  out:
> @@ -1441,6 +1440,10 @@ static bool ext4_match(struct inode *parent,
>  #if IS_ENABLED(CONFIG_UNICODE)
>  	if (parent->i_sb->s_encoding && IS_CASEFOLDED(parent) &&
>  	    (!IS_ENCRYPTED(parent) || fscrypt_has_encryption_key(parent))) {
> +		struct unicode_name u = {
> +			.folded_name = &fname->cf_name,
> +			.usr_name = fname->usr_fname
> +		};
>  		int ret;
>  
>  		if (fname->cf_name.name) {
> @@ -1452,14 +1455,9 @@ static bool ext4_match(struct inode *parent,
>  					return false;
>  				}
>  			}
> -
> -			ret = ext4_match_ci(parent, &fname->cf_name, de->name,
> -					    de->name_len, true);
> -		} else {
> -			ret = ext4_match_ci(parent, fname->usr_fname,
> -					    de->name, de->name_len, false);
>  		}
>  
> +		ret = ext4_match_ci(parent, &u, de->name, de->name_len);
>  		if (ret < 0) {
>  			/*
>  			 * Treat comparison errors as not a match.  The
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index e2d892b201b0..3f76a18a5f40 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -3358,6 +3358,11 @@ extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
>  
>  extern int generic_check_addressable(unsigned, u64);
>  
> +struct unicode_name {
> +	const struct qstr *folded_name;
> +	const struct qstr *usr_name;
> +};
> +
>  extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
>  
>  #ifdef CONFIG_MIGRATION

I don't really see the point of this.  The only times struct unicode_name gets
used are when one is initialized on the stack for a single call to
generic_ci_match().  So the end result is just that the function prototype is:

int generic_ci_match(const struct inode *parent,
		     const struct unicode_name *uname,
		     const u8 *de_name, size_t de_name_len);

... instead of:

int generic_ci_match(const struct inode *parent, const struct qstr *usr_fname,
		     const struct qstr *folded_name,
		     const u8 *de_name, size_t de_name_len);

So the only effect is to consolidate two parameters into one.  I don't think
it's worth it, given that the struct is being created on-demand.

Also note that filenames are not necessarily valid Unicode, so "unicode_name" is
a bit misleading.

- Eric