Re: [patch V2 00/18] mm/highmem: Preemptible variant of kmap_atomic & friends

From: Linus Torvalds
Date: Thu Oct 29 2020 - 19:11:38 EST


On Thu, Oct 29, 2020 at 3:32 PM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>
>
> Though I wanted to share the current state of affairs before investigating
> that further. If there is consensus in going forward with this, I'll have a
> deeper look into this issue.

Me likee. I think this looks like the right thing to do.

I didn't actually apply the patches, but just from reading them it
_looks_ to me like you do the migrate_disable() unconditionally, even
if it's not a highmem page..

That sounds like it might be a good thing for debugging, but not
necessarily great in general.

Or am I misreading things?

Linus