[6.8-rc1 Regression] Unable to exec apparmor_parser from virt-aa-helper

[John Johansen]

On 1/24/24 10:57, Kees Cook wrote:
> On Wed, Jan 24, 2024 at 09:10:58AM -0800, Linus Torvalds wrote:
>> On Wed, 24 Jan 2024 at 08:54, Linus Torvalds
>> <torvalds at linux-foundation.org> wrote:
>>>
>>> Hmm. That whole thing is disgusting. I think it should have checked
>>> FMODE_EXEC, and I have no idea why it doesn't.
>>
>> Maybe because FMODE_EXEC gets set for uselib() calls too? I dunno. I
>> think it would be even better if we had the 'intent' flags from
>> 'struct open_flags' available, but they aren't there in the
>> file_open() security chain.
> 
> I've tested AppArmor, and this works fine:
> 
thanks. I also ran it through the regression test suit, to double
check so that Murphy doesn't bite.

that this even tripped a regression is a bug that I am going to
have to chase down. The file check at this point should just be
redundant.

thanks for the quick fix

> diff --git a/security/apparmor/lsm.c b/security/apparmor/lsm.c
> index 7717354ce095..ab104ce05f96 100644
> --- a/security/apparmor/lsm.c
> +++ b/security/apparmor/lsm.c
> @@ -470,7 +470,7 @@ static int apparmor_file_open(struct file *file)
>   	 * implicit read and executable mmap which are required to
>   	 * actually execute the image.
>   	 */
> -	if (current->in_execve) {
> +	if (file->f_flags & __FMODE_EXEC) {
>   		fctx->allow = MAY_EXEC | MAY_READ | AA_EXEC_MMAP;
>   		return 0;
>   	}
> 
> Converting TOMOYO is less obvious to me, though, as it has a helper that
> isn't strictly always called during open(). I haven't finished figuring
> out the call graphs for it...
>