[PATCH 1/2] lockdep: improve current->(hard|soft)irqs_enabled synchronisation with actual irq state

[peterz at infradead.org]

On Sun, Jul 26, 2020 at 02:14:34PM +1000, Nicholas Piggin wrote:
> > Now, x86, and at least arm64 call nmi_enter() before
> > trace_hardirqs_off(), but AFAICT Power never did that, and that's part
> > of the problem. nmi_enter() does lockdep_off() and that _used_ to also
> > kill IRQ tracking.
> 
> Power does do nmi_enter -- __perf_event_interrupt()
> 
>         nmi = perf_intr_is_nmi(regs);
>         if (nmi)
>                 nmi_enter();
>         else
>                 irq_enter();

That's _waaaay_ too late, you've done the trace_hardirqs_{off,on} in
arch/powerpc/kernel/exceptions-64e.S, you need to _first_ do nmi_enter()
and _then_ trace_hardirqs_off(), or rather, that's still broken, but
then we get into the details of entry ordering.