Re: [GIT PULL] Folio patches for 5.18 (MM part)

[David Hildenbrand <david@redhat.com>]

On 21.03.22 16:31, Matthew Wilcox wrote:
> On Mon, Mar 21, 2022 at 04:24:08PM +0100, David Hildenbrand wrote:
>> On 21.03.22 16:16, David Hildenbrand wrote:
>>> On 21.03.22 14:30, Matthew Wilcox wrote:
>>>> Hi Linus,
>>>>
>>>> This is the first of two folio-related pull requests for this merge
>>>> window.  This is the MM side of things and we had some unfortunate
>>>> complex merge conflicts to resolve.  I decided to redo my changes on
>>>> top of Hugh's and Christoph's patches, so I'm the one sending the
>>>> pull request.
>>>>
>>>> The following changes since commit f71077a4d84bbe8c7b91b7db7c4ef815755ac5e3:
>>>>
>>>>   Merge tag 'mmc-v5.17-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc (2022-02-16 12:09:22 -0800)
>>>>
>>>> are available in the Git repository at:
>>>>
>>>>   git://git.infradead.org/users/willy/pagecache.git tags/folio-5.18
>>>>
>>>> for you to fetch changes up to 5063f22c914e3e5f2239cf91f4986042dc705bde:
>>>>
>>>>   mm/damon: minor cleanup for damon_pa_young (2022-03-16 10:09:50 -0400)
>>>>
>>>> ----------------------------------------------------------------
>>>> Folio changes for 5.18
>>>>
>>>> Several of us had overlapping, conflicting changes to the MM this
>>>> round, and I volunteered to send the pull request.
>>>>
>>>>  - Hugh rewrote how munlock works to massively reduce the contention
>>>>    on i_mmap_rwsem:
>>>>    https://lore.kernel.org/linux-mm/8e4356d-9622-a7f0-b2c-f116b5f2efea@google.com/
>>>>  - Christoph sorted out the page refcount mess for ZONE_DEVICE pages:
>>>>    https://lore.kernel.org/linux-mm/20220210072828.2930359-1-hch@lst.de/
>>>>  - I converted GUP to use folios and make pincount available for order-1
>>>>    pages.
>>>>  - I converted a few more truncation functions to use folios
>>>>  - I converted page_vma_mapped_walk to use PFNs instead of pages
>>>>  - I converted rmap_walk to use folios
>>>>  - I converted most of shrink_page_list() to use a folio
>>>>  - I added support for creating large folios in readahead
>>>>
>>>> ----------------------------------------------------------------
>>>> Alex Sierra (10):
>>>>       mm: add zone device coherent type memory support
>>>>       mm: add device coherent vma selection for memory migration
>>>>       mm/gup: fail get_user_pages for LONGTERM dev coherent type
>>>>       drm/amdkfd: add SPM support for SVM
>>>>       drm/amdkfd: coherent type as sys mem on migration to ram
>>>>       lib: test_hmm add ioctl to get zone device type
>>>>       lib: test_hmm add module param for zone device type
>>>>       lib: add support for device coherent type in test_hmm
>>>>       tools: update hmm-test to support device coherent type
>>>>       tools: update test_hmm script to support SP config
>>>>
>>>> Alistair Popple (2):
>>>>       mm: remove the vma check in migrate_vma_setup()
>>>>       mm/gup: migrate device coherent pages when pinning instead of failing
>>>
>>> ... I thought DEVICE_COHERENT is still under development?
>>>
>>
>> After verifying that I'm not daydreaming [1] (and realizing that I had
>> review comments to some of these patches that have not been resolved
>> yet) and also not spotting these change in your changelog above, I
>> assume this stuff was included by mistake. NACK to merging
>> DEVICE_COHERENT at this point.
>>
>> [1] https://lkml.kernel.org/r/20220310172633.9151-1-alex.sierra@amd.com
> 
> That patch ("split vm_normal_pages for LRU and non-LRU handling") isn't
> included in this pull request.  The patches I have were those sent by
> Christoph here:
> 
> https://lore.kernel.org/linux-mm/20220210072828.2930359-1-hch@lst.de/
> 
> I can drop any patches that you have objections to, but I don't see
> any objections from you to any patches in that list.

Well, I was discussing with the original authors about how to proceed.

Quoting from:

https://lkml.kernel.org/r/1747447c-202d-9195-9d44-57f299be48c4@amd.com

"
Yes, it should be part of that series. Alex developed it on top of the
series for now. But I think eventually it would need to be spliced into it.

Patch1 would need to go somewhere before the other DEVICE_COHERENT
patches (with minor modifications). Patch 2 could be squashed into
"tools: add hmm gup test for long term pinned device pages" or go next
to it. Patch 3 doesn't have a direct dependency on device-coherent
pages. It only mentions them in comments.
"

I can understand that Christoph included a rebased version in his rework
(to keep it working in -next and/or help the original authors?), but to
me that doesn't mean that the feature is finally done.

-- 
Thanks,

David / dhildenb