On Tue, Jun 23, 2020 at 05:23:26PM +0200, Peter Zijlstra wrote:
> On Tue, Jun 23, 2020 at 04:59:14PM +0200, Joerg Roedel wrote:
> > On Tue, Jun 23, 2020 at 04:53:44PM +0200, Peter Zijlstra wrote:
> > > +noinstr void idtentry_validate_ist(struct pt_regs *regs)
> > > +{
> > > + if ((regs->sp & ~(EXCEPTION_STKSZ-1)) ==
> > > + (_RET_IP_ & ~(EXCEPTION_STKSZ-1)))
> > > + die("IST stack recursion", regs, 0);
> > > +}
> >
> > Yes, this is a start, it doesn't cover the case where the NMI stack is
> > in-between, so I think you need to walk down regs->sp too.
>
> That shouldn't be possible with the current code, I think.
To clarify, we have: NMI, MCE, DB and DF.
DF (with the exception of ESPFIX) is fatal.
MCE from kernel is fatal (which is what makes the MCE in NMI 'work')
NMI and DB clear DR7, which avoids DB in NMI.
So that leaves: NMI in DB, and that works.
