Re: [PATCH 1/3] x86/entry/64: Refactor IRQ stacks and make then NMI-safe

[Linus Torvalds]

On Sat, Jul 25, 2015 at 10:59 AM, Andy Lutomirski <luto@xxxxxxxxxxxxxx> wrote:
>
> What if we added something like:
>
> if (regs->ip == ret_after_sti && !user_mode(regs) && (regs->flags &
> X86_EFLAGS_IF)) {
>   regs->ip--;
>   regs->flags &= ~X86_EFLAGS_IF;
> }
>
> to do_nmi, do_machine_check, and do_debug (the latter because kernel
> breakpoints, damnit)?

Hmm. And how would you test it?

Putting an instruction breakpoint on the final 'ret' might do it, I
guess. "mov ss" disables even that (and is documented to disable nmi
too), but maybe it works with 'sti; ret'.

But yes, as long as we'd have some test coverage, that sounds doable.

                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/