Re: [PATCH v2 4/4] seccomp: Don't special case audited processes when logging

[Kees Cook]

On Wed, May 2, 2018 at 8:53 AM, Tyler Hicks <tyhicks@xxxxxxxxxxxxx> wrote:
> diff --git a/kernel/seccomp.c b/kernel/seccomp.c
> index da78835..9029d9d 100644
> --- a/kernel/seccomp.c
> +++ b/kernel/seccomp.c
> @@ -584,18 +584,13 @@ static inline void seccomp_log(unsigned long syscall, long signr, u32 action,
>         }
>
>         /*
> -        * Force an audit message to be emitted when the action is RET_KILL_*,
> -        * RET_LOG, or the FILTER_FLAG_LOG bit was set and the action is
> -        * allowed to be logged by the admin.
> +        * Emit an audit message when the action is RET_KILL_*, RET_LOG, or the
> +        * FILTER_FLAG_LOG bit was set. The admin has the ability to silence
> +        * any action from being logged by removing the action name from the
> +        * seccomp_actions_logged sysctl.
>          */
>         if (log)
> -               return __audit_seccomp(syscall, signr, action);
> -
> -       /*
> -        * Let the audit subsystem decide if the action should be audited based
> -        * on whether the current task itself is being audited.
> -        */
> -       return audit_seccomp(syscall, signr, action);
> +               audit_seccomp(syscall, signr, action);
>  }

This whole series looks great to me. If I can get an Ack from Paul for
the audit bits, I can take it via the seccomp tree. One minor nit on
seccomp_log() above, I'd probably change this to show the "exception"
case as "out of line" of normal code flow. i.e. instead of "if (log)
audit_seccomp", invert it to return early:

...
    if (!log)
        return;

    audit_seccomp(syscall, signr, action);
}

But if there isn't some other need for a v3, I can just make this
change when I commit.

Thanks for fixing this up!

-Kees

-- 
Kees Cook
Pixel Security