Re: Smatch check for Spectre stuff

From: Oleg Nesterov
Date: Fri Apr 20 2018 - 13:21:16 EST


On 04/20, Thomas Gleixner wrote:
>
> > kernel/signal.c:3457 do_sigaction() warn: potential spectre issue 'p->sighand->action'
>
> This one is correctly detected

Not sure,

k = &p->sighand->action[sig-1];

calculates the addr, although we do '*oact = *k' later. I dunno.

> > kernel/signal.c:65 sig_handler() warn: potential spectre issue 't->sighand->action'
>
> It's unclear from a quick look whether it's really possible to speculate
> all over the other things there, but we probably err out on the safe side

at least not in do_sigaction(). And in fact do_sigaction() doesn't need
sig_handler() at all, it could use act->sa.sa_handler instead.

Oleg.