Re: Linux Kernel Defence Map

[Kees Cook <keescook () chromium org>]

On Thu, Apr 5, 2018 at 4:38 PM, Alexander Popov <alex.popov () linux com> wrote:
> On 05.04.2018 22:20, Kees Cook wrote:
> > On Thu, Apr 5, 2018 at 5:32 AM, Alexander Popov <alex.popov () linux com> wrote:
> > > On 05.04.2018 01:17, Kees Cook wrote:
> > "type confusion" seems weird to me, but I haven't spent a lot of time
> > weighing the options of the naming of these things. "Overwriting a
> > function pointer" is the method, and the bug is "unexpectedly
> > accessing userspace memory from the kernel" (which is usually
> > "something overwrite a pointer").
>
> Just got an idea to call it "userspace data access". Short and simple!
>
> I also combined SMAP/PAN and UDEREF into a cluster to reduce the number of
> edges. Now it looks a bit better.
>
> > > Kees, thanks again for such a cool feedback. The map is updated.
> >
> > Very cool! Maybe also add an out-of-tree bubble for "Clang CFI", which
> > gives forward-edge protection for code-reuse...
>
> Ok. Created a CFI cluster with RAP and Clang CFI inside.

Well, naming can get confusing here. RAP got renamed along the way to
include both "Return Address Protection" (backward edge) and "Indirect
Control Transfer Protection" (forward edge). Clang CFI is forward edge
only, though things like shadow stacks or pointer authentication can
provide backward edge protection. So... I'm not sure how much detail
you want to capture in the bubbles. :)

> However, I didn't manage to find any materials about applying Clang CFI to the
> Linux kernel.

Here's Sami Tolvanen's work on doing Clang CFI (on top of Clang LTO):
https://android-review.googlesource.com/q/topic:android-4.9-cfi

-Kees

-- 
Kees Cook
Pixel Security