Re: [PATCH 4/5] mm/memcontrol: allow to uncharge page without using page->lru field

[Balbir Singh]

On Wed, Jul 5, 2017 at 4:38 PM, Michal Hocko <mhocko@xxxxxxxxxx> wrote:
> On Wed 05-07-17 13:18:18, Balbir Singh wrote:
>> On Tue, Jul 4, 2017 at 10:51 PM, Michal Hocko <mhocko@xxxxxxxxxx> wrote:
>> > On Mon 03-07-17 17:14:14, JÃrÃme Glisse wrote:
>> >> HMM pages (private or public device pages) are ZONE_DEVICE page and
>> >> thus you can not use page->lru fields of those pages. This patch
>> >> re-arrange the uncharge to allow single page to be uncharge without
>> >> modifying the lru field of the struct page.
>> >>
>> >> There is no change to memcontrol logic, it is the same as it was
>> >> before this patch.
>> >
>> > What is the memcg semantic of the memory? Why is it even charged? AFAIR
>> > this is not a reclaimable memory. If yes how are we going to deal with
>> > memory limits? What should happen if go OOM? Does killing an process
>> > actually help to release that memory? Isn't it pinned by a device?
>> >
>> > For the patch itself. It is quite ugly but I haven't spotted anything
>> > obviously wrong with it. It is the memcg semantic with this class of
>> > memory which makes me worried.
>> >
>>
>> This is the HMM CDM case. Memory is normally malloc'd and then
>> migrated to ZONE_DEVICE or vice-versa. One of the things we did
>> discuss was seeing ZONE_DEVICE memory in user page tables.
>
> This doesn't answer any of the above questions though.


Jerome is the expert and I am sure he has a better answer, but my understanding
is that this path gets called through release_pages() <-- zap_pte_range().
At first even I pondered about the same thing, but then came across this path.

Balbir Singh.