Re: [PATCH 1/2] mem-hotplug: use GFP_HIGHUSER_MOVABLE in, alloc_migrate_target()

[Xishi Qiu]

On 2016/7/19 14:50, Joonsoo Kim wrote:

> On Mon, Jul 18, 2016 at 04:31:11PM +0800, Xishi Qiu wrote:
>> On 2016/7/18 16:05, Vlastimil Babka wrote:
>>
>>> On 07/18/2016 10:00 AM, Xishi Qiu wrote:
>>>> On 2016/7/18 13:51, Joonsoo Kim wrote:
>>>>
>>>>> On Fri, Jul 15, 2016 at 10:47:06AM +0800, Xishi Qiu wrote:
>>>>>> alloc_migrate_target() is called from migrate_pages(), and the page
>>>>>> is always from user space, so we can add __GFP_HIGHMEM directly.
>>>>>
>>>>> No, all migratable pages are not from user space. For example,
>>>>> blockdev file cache has __GFP_MOVABLE and migratable but it has no
>>>>> __GFP_HIGHMEM and __GFP_USER.
>>>>>
>>>>
>>>> Hi Joonsoo,
>>>>
>>>> So the original code "gfp_t gfp_mask = GFP_USER | __GFP_MOVABLE;"
>>>> is not correct?
>>>
>>> It's not incorrect. GFP_USER just specifies some reclaim flags, and may perhaps restrict allocation through __GFP_HARDWALL, where the original
>>> page could have been allocated without the restriction. But it doesn't put the place in an unexpected address range, as placing a non-highmem page into highmem could. __GFP_MOVABLE then just controls a heuristic for placement within a zone.
>>>
>>>>> And, zram's memory isn't GFP_HIGHUSER_MOVABLE but has __GFP_MOVABLE.
>>>>>
>>>>
>>>> Can we distinguish __GFP_MOVABLE or GFP_HIGHUSER_MOVABLE when doing
>>>> mem-hotplug?
>>>
>>> I don't understand the question here, can you rephrase with more detail? Thanks.
>>>
>>
>> Hi Joonsoo,
> 
> Above is answered by Vlastimil. :)
> 
>> When we do memory offline, and the zone is movable zone,
>> can we use "alloc_pages_node(nid, GFP_HIGHUSER_MOVABLE, 0);" to alloc a
>> new page? the nid is the next node.
> 
> I don't know much about memory offline, but, AFAIK, memory offline
> could happen on non-movable zone like as ZONE_NORMAL. Perhaps, you can add
> "if zone of the page is movable zone then alloc with GFP_HIGHUSER_MOVABLE".
> 
> Thanks.
> 

Hi Joonsoo and Vlastimil,

Memory offline could happen on both movable zone and non-movable zone, and we
can offline the whole node if the zone is movable_zone(the node only has one
movable_zone), and if the zone is normal_zone, we cannot offline the whole node,
because some kernel memory can't be migrated.

So how about change alloc_migrate_target() to alloc memory from the next node
with GFP_HIGHUSER_MOVABLE, if the offline zone is movable_zone.

And if the offline zone is normal_zone, we don't change anything, that means
the new page may be from the same node.

Thanks,
Xishi Qiu

> .
>