Re: [PATCH v2] signal: factor out the kernel reserved si_code check

[Oleg Nesterov <oleg@redhat.com>]

On 08/06, Bradley Morgan wrote:
>
> The check that prevents userspace from sending siginfo with si_code
> values reserved to the kernel is duplicated across do_rt_sigqueueinfo(),
> do_rt_tgsigqueueinfo() and do_pidfd_send_signal(). Move the check
> into a helper so the rule lives in one place.

> +/*
> + * Not even root can pretend to send SI_FROMKERNEL() signals.
> + * Nor can they impersonate kill()/tgkill(), which have si_pid/uid
> + */
> +static bool si_code_reserved_to_kernel(int si_code)
> +{
> +	return si_code >= 0 || si_code == SI_TKILL;
> +}

OK, Acked-by: Oleg Nesterov <oleg@redhat.com>


But can't resist... We already have SI_FROMUSER() and SI_FROMKERNEL()
both take the pointer to siginfo as an argument.

Now we also have si_code_reserved_to_kernel() which falls into the same
category, but it aceepts .si_code and the names don't match...

OK, nevermind. I can't suggest a better name. Even if we change the
new helper to take "kernel_siginfo_t *info" to match SI_FROMXXX macros.

Oleg.