Re: [PATCH v6 01/10] mm: add zone device coherent type memory support

[David Hildenbrand <david@redhat.com>]

On 11.02.22 18:07, Felix Kuehling wrote:
> 
> Am 2022-02-11 um 11:39 schrieb David Hildenbrand:
>> On 11.02.22 17:15, David Hildenbrand wrote:
>>> On 01.02.22 16:48, Alex Sierra wrote:
>>>> Device memory that is cache coherent from device and CPU point of view.
>>>> This is used on platforms that have an advanced system bus (like CAPI
>>>> or CXL). Any page of a process can be migrated to such memory. However,
>>>> no one should be allowed to pin such memory so that it can always be
>>>> evicted.
>>>>
>>>> Signed-off-by: Alex Sierra <alex.sierra@amd.com>
>>>> Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
>>>> Reviewed-by: Alistair Popple <apopple@nvidia.com>
>>> So, I'm currently messing with PageAnon() pages and CoW semantics ...
>>> all these PageAnon() ZONE_DEVICE variants don't necessarily make my life
>>> easier but I'm not sure yet if they make my life harder. I hope you can
>>> help me understand some of that stuff.
>>>
>>> 1) What are expected CoW semantics for DEVICE_COHERENT?
>>>
>>> I assume we'll share them just like other PageAnon() pages during fork()
>>> readable, and the first sharer writing to them receives an "ordinary"
>>> !ZONE_DEVICE copy.
>>>
>>> So this would be just like DEVICE_EXCLUSIVE CoW handling I assume, just
>>> that we don't have to go through the loop of restoring a device
>>> exclusive entry?
>>>
>>> 2) How are these pages freed to clear/invalidate PageAnon() ?
>>>
>>> I assume for PageAnon() ZONE_DEVICE pages we'll always for via
>>> free_devmap_managed_page(), correct?
>>>
>>>
>>> 3) FOLL_PIN
>>>
>>> While you write "no one should be allowed to pin such memory", patch #2
>>> only blocks FOLL_LONGTERM. So I assume we allow ordinary FOLL_PIN and
>>> you might want to be a bit more precise?
>>>
>>>
>>> ... I'm pretty sure we cannot FOLL_PIN DEVICE_PRIVATE pages, but can we
>>> FILL_PIN DEVICE_EXCLUSIVE pages? I strongly assume so?
>>>
>>>
>>> Thanks for any information.
>>>
>> (digging a bit more, I realized that device exclusive pages are not
>> actually/necessarily ZONE_DEVICE pages -- so I assume DEVICE_COHERENT
>> will be the actual first PageAnon() ZONE_DEVICE pages that can be
>> present in a page table.)
> 
> I think DEVICE_GENERIC pages can also be mapped in the page table. In 
> fact, the first version of our patches attempted to add migration 
> support to DEVICE_GENERIC. But we were convinced to create a new 
> ZONE_DEVICE page type for our use case instead.

Do you know if DEVICE_GENERIC pages would end up as PageAnon()? My
assumption was that they would be part of a special mapping.

-- 
Thanks,

David / dhildenb