Re: [PATCH V5 0/3] arm64/mm: Enable memory hot remove

From: David Hildenbrand
Date: Thu May 30 2019 - 03:26:49 EST


On 30.05.19 06:23, Anshuman Khandual wrote:
>
>
> On 05/30/2019 03:36 AM, Andrew Morton wrote:
>> On Wed, 29 May 2019 14:46:24 +0530 Anshuman Khandual <anshuman.khandual@xxxxxxx> wrote:
>>
>>> This series enables memory hot remove on arm64 after fixing a memblock
>>> removal ordering problem in generic __remove_memory() and one possible
>>> arm64 platform specific kernel page table race condition. This series
>>> is based on latest v5.2-rc2 tag.
>>
>> Unfortunately this series clashes syntactically and semantically with
>> David Hildenbrand's series "mm/memory_hotplug: Factor out memory block
>> devicehandling". Could you and David please figure out what we should
>> do here?
>>
>
> Hello Andrew,
>
> I was able to apply the above mentioned V3 series [1] from David with some changes
> listed below which tests positively on arm64. These changes assume that the arm64
> hot-remove series (current V5) gets applied first.
>
> Changes to David's series
>
> A) Please drop (https://patchwork.kernel.org/patch/10962565/) [v3,04/11]
>
> - arch_remove_memory() is already being added through hot-remove series
>
> B) Rebase (https://patchwork.kernel.org/patch/10962575/) [v3, 06/11]
>
> - arm64 hot-remove series adds CONFIG_MEMORY_HOTREMOVE wrapper around
> arch_remove_memory() which can be dropped in the rebased patch
>
> C) Rebase (https://patchwork.kernel.org/patch/10962589/) [v3, 09/11]
>
> - hot-remove series moves arch_remove_memory() before memblock_[free|remove]()
> - So remove_memory_block_devices() should be moved before arch_remove_memory()
> in it's new position
>
> David,
>
> Please do let me know if the plan sounds good or you have some other suggestions.

That's exactly what I had in mind :)

Andrew, you can drop my series and pick up Anshumans series first. I can
then rebase and resend.
Cheers!

--

Thanks,

David / dhildenb