Re: [PATCH v11 25/25] x86/cet/shstk: Add arch_prctl functions for shadow stack

[Yu, Yu-cheng]

On 9/16/2020 6:52 AM, Andy Lutomirski wrote:
> On Mon, Sep 14, 2020 at 2:14 PM Dave Hansen <dave.hansen@xxxxxxxxx> wrote:
> > On 9/14/20 11:31 AM, Andy Lutomirski wrote:
> > > No matter what we do, the effects of calling vfork() are going to be a
> > > bit odd with SHSTK enabled.  I suppose we could disallow this, but
> > > that seems likely to cause its own issues.
> >
> > What's odd about it?  If you're a vfork()'d child, you can't touch the
> > stack at all, right?  If you do, you or your parent will probably die a
> > horrible death.
>
> An evil program could vfork(), have the child do a bunch of returns
> and a bunch of calls, and exit.  The net effect would be to change the
> parent's shadow stack contents.  In a sufficiently strict model, this
> is potentially problematic.

When a vfork child returns, its parent's shadow stack pointer is where 
it was before the child starts.  To move the shadow stack pointer and 
re-use the content left by the child, the parent needs to use CALL, RET, 
INCSSP, or RSTORSSP.  This seems to be difficult.

> The question is: how much do we want to protect userspace from itself?
>

If any issue comes up, people can always find ways to counter it.

> --Andy