Re: [patch 1/2] x86_64 page fault NMI-safe

[Linus Torvalds]

On Wed, Jul 14, 2010 at 3:51 PM, Jeremy Fitzhardinge <jeremy@xxxxxxxx> wrote:
>
> We screw around with iret because there's a separate interrupt mask flag
> which can't be set atomically with respect to a stack/ring change (well,
> there's more to it, but I won't confuse matters).

Umm, I know. It's what this whole discussion (non-paravirtualized) is
all about. And I have a suggestion that should fix the
non-paravirtualized case _without_ actually touching anything but the
NMI code itself.

What I tried to say is that the paravirtualized people should take a
look at my suggestion, and see if they can apply the logic to their
NMI handling too. And in the process totally remove the need for
paravirtualizing iret, exactly because the approach handles the magic
NMI lock logic entirely in the NMI handler itself.

Because I think the same thing that would make us not need to worry
about nested page faults _during_ NMI (because we could make the NMI
code do the right thing even when the hardware doesn't lock out NMI's
for us) is also the exact same logic that the paravirt monitor could
do for its own NMI handling.

Wouldn't it be nice to be able to remove the need to paravirtualize iret?

                  Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/