linux-kernel - Re: [PATCH] fsnotify, lsm: Separate fsnotify_open_perm() and security_file

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] [day] [month] [year] [list]

Message-ID: <20241014141153.swwzlqa22ndkzbhn@quack3>
Date: Mon, 14 Oct 2024 16:11:53 +0200
From: Jan Kara <jack@...e.cz>
To: Song Liu <song@...nel.org>
Cc: linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-security-module@...r.kernel.org, viro@...iv.linux.org.uk,
	brauner@...nel.org, jack@...e.cz, paul@...l-moore.com,
	jmorris@...ei.org, serge@...lyn.com, kernel-team@...com
Subject: Re: [PATCH] fsnotify, lsm: Separate fsnotify_open_perm() and
 security_file_open()

On Fri 11-10-24 13:37:22, Song Liu wrote:
> Currently, fsnotify_open_perm() is called from security_file_open(). This
> is not right for CONFIG_SECURITY=n and CONFIG_FSNOTIFY=y case, as
> security_file_open() in this combination will be a no-op and not call
> fsnotify_open_perm(). Fix this by calling fsnotify_open_perm() directly.
> 
> Signed-off-by: Song Liu <song@...nel.org>
> 
> ---
> 
> PS: I didn't included a Fixes tag. This issue was probably introduced 15
> years ago in [1]. If we want to back port this to stable, we will need
> another version for older kernel because of [2].

Well, this is not a problem because CONFIG_FANOTIFY_ACCESS_PERMISSIONS has
"depends on SECURITY". So fsnotify_open_perm() can have anything to do only
if CONFIG_SECURITY is enabled. That being said I agree it makes sense to
decouple security & fsnotify because there's no significant benefit and
only confusion. So I like the patch but please update the changelog and
also finish the split as Amir suggests.

								Honza

> [1] c4ec54b40d33 ("fsnotify: new fsnotify hooks and events types for access decisions")
> [2] 36e28c42187c ("fsnotify: split fsnotify_perm() into two hooks")
> ---
>  fs/open.c           | 4 ++++
>  security/security.c | 9 +--------
>  2 files changed, 5 insertions(+), 8 deletions(-)
> 
> diff --git a/fs/open.c b/fs/open.c
> index acaeb3e25c88..6c4950f19cfb 100644
> --- a/fs/open.c
> +++ b/fs/open.c
> @@ -946,6 +946,10 @@ static int do_dentry_open(struct file *f,
>  	if (error)
>  		goto cleanup_all;
>  
> +	error = fsnotify_open_perm(f);
> +	if (error)
> +		goto cleanup_all;
> +
>  	error = break_lease(file_inode(f), f->f_flags);
>  	if (error)
>  		goto cleanup_all;
> diff --git a/security/security.c b/security/security.c
> index 6875eb4a59fc..a72cc62c0a07 100644
> --- a/security/security.c
> +++ b/security/security.c
> @@ -19,7 +19,6 @@
>  #include <linux/kernel.h>
>  #include <linux/kernel_read_file.h>
>  #include <linux/lsm_hooks.h>
> -#include <linux/fsnotify.h>
>  #include <linux/mman.h>
>  #include <linux/mount.h>
>  #include <linux/personality.h>
> @@ -3102,13 +3101,7 @@ int security_file_receive(struct file *file)
>   */
>  int security_file_open(struct file *file)
>  {
> -	int ret;
> -
> -	ret = call_int_hook(file_open, file);
> -	if (ret)
> -		return ret;
> -
> -	return fsnotify_open_perm(file);
> +	return call_int_hook(file_open, file);
>  }
>  
>  /**
> -- 
> 2.43.5
> 
-- 
Jan Kara <jack@...e.com>
SUSE Labs, CR