Re: [PATCH] apparmor: don't try to replace stale label in ptraceme check

[John Johansen]

On 09/28/2018 06:49 PM, Jann Horn wrote:
> begin_current_label_crit_section() must run in sleepable context because
> when label_is_stale() is true, aa_replace_current_label() runs, which uses
> prepare_creds(), which can sleep.
> 
> Until now, the ptraceme access check (which runs with tasklist_lock held)
> violated this rule.
> 
yep, thanks

I've pulled this into my tree and will send it up with another fix

Acked-by: John Johansen <john.johansen@xxxxxxxxxxxxx>

> Fixes: b2d09ae449ced ("apparmor: move ptrace checks to using labels")
> Reported-by: Cyrill Gorcunov <gorcunov@xxxxxxxxx>
> Reported-by: kernel test robot <rong.a.chen@xxxxxxxxx>
> Signed-off-by: Jann Horn <jannh@xxxxxxxxxx>
> ---
>  security/apparmor/lsm.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/security/apparmor/lsm.c b/security/apparmor/lsm.c
> index 8c7f46a6a8dc..0f56431b4b2f 100644
> --- a/security/apparmor/lsm.c
> +++ b/security/apparmor/lsm.c
> @@ -131,11 +131,11 @@ static int apparmor_ptrace_traceme(struct task_struct *parent)
>  	struct aa_label *tracer, *tracee;
>  	int error;
>  
> -	tracee = begin_current_label_crit_section();
> +	tracee = __begin_current_label_crit_section();
>  	tracer = aa_get_task_label(parent);
>  	error = aa_may_ptrace(tracer, tracee, AA_PTRACE_TRACE);
>  	aa_put_label(tracer);
> -	end_current_label_crit_section(tracee);
> +	__end_current_label_crit_section(tracee);
>  
>  	return error;
>  }
>