Re: [PATCH 06/18] add support for Clang's Shadow Call Stack (SCS)

[Sami Tolvanen]

On Tue, Oct 22, 2019 at 9:28 AM Mark Rutland <mark.rutland@xxxxxxx> wrote:
> I think it would be preferable to follow the example of CC_FLAGS_FTRACE
> so that this can be filtered out, e.g.
>
> ifdef CONFIG_SHADOW_CALL_STACK
> CFLAGS_SCS := -fsanitize=shadow-call-stack
> KBUILD_CFLAGS += $(CFLAGS_SCS)
> export CC_FLAGS_SCS
> endif

Sure, SGTM.

> > +choice
> > +     prompt "Return-oriented programming (ROP) protection"
> > +     default ROP_PROTECTION_NONE
> > +     help
> > +       This option controls kernel protections against return-oriented
> > +       programming (ROP) attacks.
>
> Are we expecting more options here in future?

Yes, I believe we'd be interested in seeing PAC support too once
hardware is more readily available.

> I think it would be better to ./make that depend on !SHADOW_CALL_STACK, as
> it's plausible that we can add a different ROP protection mechanism that
> is compatible with kretprobes.

OK, I can change that and remove the choice. We can always add it back
when other alternatives are added.

> > +config SHADOW_CALL_STACK
> > +     bool "Clang Shadow Call Stack"
> > +     depends on ARCH_SUPPORTS_SHADOW_CALL_STACK
> > +     depends on CC_IS_CLANG && CLANG_VERSION >= 70000
>
> Is there a reason for an explicit version check rather than a
> CC_HAS_<feature> check? e.g. was this available but broken in prior
> versions of clang?

No, this feature was added in Clang 7. However,
-fsanitize=shadow-call-stack might require architecture-specific
flags, so a simple $(cc-option, -fsanitize=shadow-call-stack) in
arch/Kconfig is not going to work. I could add something like this to
arch/arm64/Kconfig though:

select ARCH_SUPPORTS_SHADOW_CALL_STACK if CC_HAVE_SHADOW_CALL_STACK
...
config CC_HAVE_SHADOW_CALL_STACK
       def_bool $(cc-option, -fsanitize=shadow-call-stack -ffixed-x18)

And then drop CC_IS_CLANG and version check entirely. Thoughts?

> > +#define SCS_GFP                      (GFP_KERNEL | __GFP_ZERO)
>
> Normally GFP_ is a prefix. For consistency, GFP_SCS would be preferable.

Ack.

> > +extern unsigned long init_shadow_call_stack[];
>
> Do we need this exposed here? IIUC this is only assigned by assembly in
> arch code.

True, it's not needed.

> [...]
>
> > +void scs_set_init_magic(struct task_struct *tsk)
> > +{
> > +     scs_save(tsk);
> > +     scs_set_magic(tsk);
> > +     scs_load(tsk);
> > +}
>
> Can we initialize this at compile time instead?

We can. I'll change this and drop the function.


Sami