Re: [PATCH v2] mm/migrate: fix softlockup in do_pages_move() with huge page arrays

[David Hildenbrand (Arm)]

On 11/9/26 13:33, shao.mingyin@xxxxxxxxxx wrote:
> Hi David,
> 
> fair enough -- and your reading is correct: preempt=none is unselectable
> under PREEMPT_DYNAMIC, and the arches that can still pick PREEMPT_NONE
> have no NUMA, so mainline cannot reach this path.
> 
> I was aiming at the 6.6 LTS configs that shipped PREEMPT_NONE + NUMA: two
> lines of cond_resched(), no behaviour change, same as do_pages_stat() got
> in 8e934ab44389.  If you and Andrew would rather not carry a stable-only
> fix, just say so and I will withdraw it.

I think this one will be the last ones where we do that and don't really have a
real workload run into it.

It's not a good use of our time and it doesn't make upstream any better :)

-- 
Cheers,

David