On Tue, Dec 15, 2020 at 06:09:02PM -0800, Andy Lutomirski wrote:
> On Tue, Dec 15, 2020 at 5:32 PM Ira Weiny <ira.weiny@xxxxxxxxx> wrote:
> >
> > On Fri, Dec 11, 2020 at 02:14:28PM -0800, Andy Lutomirski wrote:
> > > On Mon, Nov 23, 2020 at 10:10 PM <ira.weiny@xxxxxxxxx> wrote:
>
> > > IOW we have:
> > >
> > > struct extended_pt_regs {
> > > bool rcu_whatever;
> > > other generic fields here;
> > > struct arch_extended_pt_regs arch_regs;
> > > struct pt_regs regs;
> > > };
> > >
> > > and arch_extended_pt_regs has unsigned long pks;
> > >
> > > and instead of passing a pointer to irqentry_state_t to the generic
> > > entry/exit code, we just pass a pt_regs pointer. And we have a little
> > > accessor like:
> > >
> > > struct extended_pt_regs *extended_regs(struct pt_regs *) { return
> > > container_of(...); }
> > >
> > > And we tell eBPF that extended_pt_regs is NOT ABI, and we will change
> > > it whenever we feel like just to keep you on your toes, thank you very
> > > much.
> > >
> > > Does this seem reasonable?
> >
> > Conceptually yes. But I'm failing to see how this implementation can be made
> > generic for the generic fields. The pks fields, assuming they stay x86
> > specific, would be reasonable to add in PUSH_AND_CLEAR_REGS. But the
> > rcu/lockdep field is generic. Wouldn't we have to modify every architecture to
> > add space for the rcu/lockdep bool?
> >
> > If not, where is a generic place that could be done? Basically I'm missing how
> > the effective stack structure can look like this:
> >
> > > struct extended_pt_regs {
> > > bool rcu_whatever;
> > > other generic fields here;
> > > struct arch_extended_pt_regs arch_regs;
> > > struct pt_regs regs;
> > > };
> >
> > It seems more reasonable to make it look like:
> >
> > #ifdef CONFIG_ARCH_HAS_SUPERVISOR_PKEYS
> > struct extended_pt_regs {
> > unsigned long pkrs;
> > struct pt_regs regs;
> > };
> > #endif
> >
> > And leave the rcu/lockdep bool passed by value as before (still in C).
>
> We could certainly do this,
I'm going to start with this basic support. Because I have 0 experience in
most of these architectures.
> but we could also allocate some generic
> space. PUSH_AND_CLEAR_REGS would get an extra instruction like:
>
> subq %rsp, $GENERIC_PTREGS_SIZE
>
> or however this should be written. That field would be defined in
> asm-offsets.c. And yes, all the generic-entry architectures would
> need to get onboard.
What do you mean by 'generic-entry' architectures? I thought they all used the
generic entry code?
Regardless I would need to start another thread on this topic with any of those
architecture maintainers to see what the work load would be for this. I don't
think I can do it on my own.
FWIW I think it is a bit unfair to hold up the PKS support in x86 for making
these generic fields part of the stack frame. So perhaps that could be made a
follow on to the PKS series?
>
> If we wanted to be fancy, we could split the generic area into
> initialize-to-zero and uninitialized for debugging purposes, but that
> might be more complication than is worthwhile.
Ok, agreed, but this is step 3 or 4 at the earliest.
Ira
