Re: [PATCH 1/3] x86/mm/pat: Disable preemption around __flush_tlb_all()

From: 'Sebastian Andrzej Siewior'
Date: Wed Oct 17 2018 - 12:00:58 EST


On 2018-10-17 17:47:07 [+0200], Peter Zijlstra wrote:
> > > Ideally we'd never ever merge anything that only does local invalidates,
> > > on a global address space, that's just broken.
> >
> > A little bit late to lament about that.
>
> For this, yes :/ But for future stuff we should really not allow such
> things anymore.

so we stay as is?

> > So should we just replace it with cpa_flush_all() ?
>
> The comment there suggests that will deadlock, supposedly because the
> kernel_map_page() call can happen with IRQs disabled or such.
>
> I've not deeply looked at this.

free_pages() / __free_pages() for instance ends may end up in
kernel_map_pages() (via free_pages_prepare()). And if this is invoked
with disabled interruptsâ

Sebastian