Re: [PATCH] x86/crypto: Add missing RETs

From: Josh Poimboeuf
Date: Fri Jul 06 2018 - 10:06:55 EST

On Thu, Jul 05, 2018 at 09:58:15AM +0200, Ingo Molnar wrote:
> * Josh Poimboeuf <jpoimboe@xxxxxxxxxx> wrote:
> > > So that's still incomplete in that doesn't analyze the 32-bit build yet, right?
> >
> > We could do INT3s on 64-bit and NOPs on 32-bit.
> >
> > Or, possibly even better, we could just keep NOPs everywhere and instead
> > make objtool smart enough to detect function fallthroughs. That should
> > be pretty easy, actually. It already does it for C files.
> >
> > Something like the below should work, though it's still got a few
> > issues:
> >
> > a) objtool is currently disabled for crypto code because it doesn't
> > yet understand crypto stack re-alignments (which really needs
> > fixing anyway); and
> >
> > b) it complains about the blank xen hypercalls falling through. Those
> > aren't actual functions anyway, so we should probably annotate
> > those somehow so that objtool ignores them anyway.
> >
> > I'm a bit swamped at the moment but I can fix those once I get a little
> > more bandwidth. I at least verified that this patch caught the crypto
> > missing RETs.
> Great, I'd be perfectly fine with such an approach.
> Also, if we have that then we could re-apply Alexey's patch and switch to INT3
> (only on 64-bit kernels) without any trouble, because objtool should detect any
> execution flow bugs before the INT3 could trigger, right?
> I.e. any INT3 fault would show a combination of *both* an objtool bug and a
> probable code flow bug - which I suspect would warrant crashing the box ...

Sounds good to me. I can take Alexey's patch and submit a 64-bit
version of it, along with the relevant objtool changes (though it may
still be a few weeks before I get the chance).