Re: [PATCH 04/10] x86/cet: Handle thread shadow stack

[Yu-cheng Yu]

On Fri, 2018-06-08 at 08:01 -0700, Andy Lutomirski wrote:
> On Fri, Jun 8, 2018 at 7:53 AM Florian Weimer <fweimer@xxxxxxxxxx> wrote:
> >
> > On 06/07/2018 10:53 PM, Andy Lutomirski wrote:
> > > On Thu, Jun 7, 2018 at 12:47 PM Florian Weimer <fweimer@xxxxxxxxxx> wrote:
> > >>
> > >> On 06/07/2018 08:21 PM, Andy Lutomirski wrote:
> > >>> On Thu, Jun 7, 2018 at 7:41 AM Yu-cheng Yu <yu-cheng.yu@xxxxxxxxx> wrote:
> > >>>>
> > >>>> When fork() specifies CLONE_VM but not CLONE_VFORK, the child
> > >>>> needs a separate program stack and a separate shadow stack.
> > >>>> This patch handles allocation and freeing of the thread shadow
> > >>>> stack.
> > >>>
> > >>> Aha -- you're trying to make this automatic.  I'm not convinced this
> > >>> is a good idea.  The Linux kernel has a long and storied history of
> > >>> enabling new hardware features in ways that are almost entirely
> > >>> useless for userspace.
> > >>>
> > >>> Florian, do you have any thoughts on how the user/kernel interaction
> > >>> for the shadow stack should work?
> > >>
> > >> I have not looked at this in detail, have not played with the emulator,
> > >> and have not been privy to any discussions before these patches have
> > >> been posted, however â
> > >>
> > >> I believe that we want as little code in userspace for shadow stack
> > >> management as possible.  One concern I have is that even with the code
> > >> we arguably need for various kinds of stack unwinding, we might have
> > >> unwittingly built a generic trampoline that leads to full CET bypass.
> > >
> > > I was imagining an API like "allocate a shadow stack for the current
> > > thread, fail if the current thread already has one, and turn on the
> > > shadow stack".  glibc would call clone and then call this ABI pretty
> > > much immediately (i.e. before making any calls from which it expects
> > > to return).
> >
> > Ahh.  So you propose not to enable shadow stack enforcement on the new
> > thread even if it is enabled for the current thread?  For the cases
> > where CLONE_VM is involved?
> >
> > It will still need a new assembler wrapper which sets up the shadow
> > stack, and it's probably required to disable signals.
> >
> > I think it should be reasonable safe and actually implementable.  But
> > the benefits are not immediately obvious to me.
> 
> Doing it this way would have been my first incliniation.  It would
> avoid all the oddities of the kernel magically creating a VMA when
> clone() is called, guessing the shadow stack size, etc.  But I'm okay
> with having the kernel do it automatically, too.

HJ wanted to add a arch_prctl that allocates a new shadow stack and
switches to it.  That was mainly for swapcontext.  Perhaps we can also
use that for threads?  HJ, can you comment on this?

> I think it would be
> very nice to have a way for user code to find out the size of the
> shadow stack and change it, though.  (And relocate it, but maybe
> that's impossible.  The CET documentation doesn't have a clear
> description of the shadow stack layout.)

The shadow stack is vm_mmap'ed from memory and does not have any special
layout.  We can add a arch_prctl to find out shadow stack's address and
size.

> >
> > > We definitely want strong enough user control that tools like CRIU can
> > > continue to work.  I haven't looked at the SDM recently enough to
> > > remember for sure, but I'm reasonably confident that user code can
> > > learn the address of its own shadow stack.  If nothing else, CRIU
> > > needs to be able to restore from a context where there's a signal on
> > > the stack and the signal frame contains a shadow stack pointer.
> >
> > CRIU also needs the shadow stack *contents*, which shouldn't be directly
> > available to the process.  So it needs very special interfaces anyway.
> 
> True.  I proposed in a different email that ptrace() have full control
> of the shadow stack (read, write, lock, unlock, etc).

PTRACE can do PTRACE_POKEDATA on shadow stack.  We can add lock/unlock.

> >
> > Does CRIU implement MPX support?
> 
> Dunno.  But given that MPX seems to be dying, I'm not sure it matters.
> 
> --Andy