Re: [RFC PATCH] UML: add support for KASAN under x86_64

[Dmitry Vyukov]

On Thu, Jan 16, 2020 at 9:03 AM Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote:
>
> On Thu, 2020-01-16 at 08:57 +0100, Johannes Berg wrote:
> >
> > And if I remember from looking at KASAN, some of the constructors there
> > depended on initializing after the KASAN data structures were set up (or
> > at least allocated)? It may be that you solved that by allocating the
> > shadow so very early though.
>
> Actually, no ... it's still after main(), and the constructors run
> before.
>
> So I _think_ with the CONFIG_CONSTRUCTORS revert, this will no longer
> work (but happy to be proven wrong!), if so then I guess we do have to
> find a way to initialize the KASAN things from another (somehow
> earlier?) constructor ...
>
> Or find a way to fix CONFIG_CONSTRUCTORS and not revert, but I looked at
> it quite a bit and didn't.

Looking at this problem and at the number of KASAN_SANITIZE := n in
Makefiles (some of which are pretty sad, e.g. ignoring string.c,
kstrtox.c, vsprintf.c -- that's where the bugs are!), I think we
initialize KASAN too late. I think we need to do roughly what we do in
user-space asan (because it is user-space asan!). Constructors run
before main and it's really good, we need to initialize KASAN from
these constructors. Or if that's not enough in all cases, also add own
constructor/.preinit array entry to initialize as early as possible.
All we need to do is to call mmap syscall, there is really no
dependencies on anything kernel-related.
This should resolve the problem with constructors (after they
initialize KASAN, they can proceed to do anything they need) and it
should get rid of most KASAN_SANITIZE (in particular, all of
lib/Makefile and kernel/Makefile) and should fix stack instrumentation
(in case it does not work now). The only tiny bit we should not
instrument is the path from constructor up to mmap call.