Re: [PATCH v1] mm: migrate: don't rely on PageMovable() of newpage after unlocking it

[David Hildenbrand]

On 29.01.19 08:18, Michal Hocko wrote:
> On Mon 28-01-19 22:09:14, David Hildenbrand wrote:
>> On 28.01.19 21:19, Michal Hocko wrote:
> [...]
>>> David, could you reformulate the changelog accordingly please? My ack
>>> still holds.
>>
>> You mean reformulating + resending for stable kernels only?
> 
> I would merge your patch even if it doesn't fix any real problem _now_.
> If for not other reasons it makes the code less subtle because we no
> longer depend on this crazy __PageMovable is special. If the movable
> flag is supposed to be synchronized with the page lock then do not do
> tricks and make code more robust because the next time somebody would
> like to fix up the current semantic he might reintroduce the bug easily.
> 

Sounds good to me, I'll rephrase and resend. I'll also have a look if I
can easily refactor the whole PageMovable logic or if this is more involved.

Thanks!

-- 

Thanks,

David / dhildenb