Re: TLB flushes on fixmap changes

[Jann Horn]

On Mon, Aug 27, 2018 at 12:11 AM Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>
> On Sun, Aug 26, 2018 at 1:15 PM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
> > On Sun, 26 Aug 2018, Andy Lutomirski wrote:
> >> > On Aug 26, 2018, at 9:47 AM, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
> >> >> On Sun, Aug 26, 2018 at 7:20 AM, Andy Lutomirski <luto@xxxxxxxxxxxxxx> wrote:
> >> >>> I tried to convince Ingo to use this method for doing "write rarely"
> >> >>> and he soundly rejected it. :) I've always liked this because AFAICT,
> >> >>> it's local to the CPU. I had proposed it in
> >> >>> https://git.kernel.org/pub/scm/linux/kernel/git/kees/linux.git/commit/?h=kspp/write-rarely&id=9ab0cb2618ebbc51f830ceaa06b7d2182fe1a52d
> >> >>
> >> >> Ingo, can you clarify why you hate it?  I personally would rather use CR3, but CR0 seems like a fine first step, at least for text_poke.
> >> >
> >> > Sorry, it looks like it was tglx, not Ingo:
> >> >
> >> > https://lkml.kernel.org/r/alpine.DEB.2.20.1704071048360.1716@nanos
> >> >
> >> > This thread is long, and one thing that I think went unanswered was
> >> > "why do we want this to be fast?" the answer is: for doing page table
> >> > updates. Page tables are becoming a bigger target for attacks now, and
> >> > it's be nice if they could stay read-only unless they're getting
> >> > updated (with something like this).
> >> >
> >> >
> >> It kind of sounds like tglx would prefer the CR3 approach. And indeed my
> >> patch has a serious problem wrt the NMI code.
> >
> > That's exactly the problem I have with CR0. It leaves everything and some
> > more writeable for any code which can interrupt that section.
>
> I thought the point was that the implementation I suggested was
> NMI-proof? (And in reading Documentation/preempt-locking.txt it sounds
> like disabling interrupts is redundant to preempt_disable()? But I
> don't understand how; it looks like the preempt stuff is advisory?)

Where are you dealing with NMIs? local_irq_disable() disables the
interrupt flag, but Non-Maskable Interrupts can still come in. As far
as I know, the only way to block those is to artificially generate an
NMI yourself (Xen does that sometimes). Otherwise, you have to twiddle
CR0.WP in the NMI handler entry/exit code.