Re: [PATCH v6 02/13] mm: remove extra ZONE_DEVICE struct page refcount

[Felix Kuehling <felix.kuehling@amd.com>]

Am 2021-08-19 um 2:00 p.m. schrieb Sierra Guiza, Alejandro (Alex):
>
> On 8/18/2021 2:28 PM, Ralph Campbell wrote:
>> On 8/17/21 5:35 PM, Felix Kuehling wrote:
>>> Am 2021-08-17 um 8:01 p.m. schrieb Ralph Campbell:
>>>> On 8/12/21 11:31 PM, Alex Sierra wrote:
>>>>> From: Ralph Campbell <rcampbell@nvidia.com>
>>>>>
>>>>> ZONE_DEVICE struct pages have an extra reference count that
>>>>> complicates the
>>>>> code for put_page() and several places in the kernel that need to
>>>>> check the
>>>>> reference count to see that a page is not being used (gup,
>>>>> compaction,
>>>>> migration, etc.). Clean up the code so the reference count doesn't
>>>>> need to
>>>>> be treated specially for ZONE_DEVICE.
>>>>>
>>>>> v2:
>>>>> AS: merged this patch in linux 5.11 version
>>>>>
>>>>> v5:
>>>>> AS: add condition at try_grab_page to check for the zone device type,
>>>>> while
>>>>> page ref counter is checked less/equal to zero. In case of device
>>>>> zone, pages
>>>>> ref counter are initialized to zero.
>>>>>
>>>>> Signed-off-by: Ralph Campbell <rcampbell@nvidia.com>
>>>>> Signed-off-by: Alex Sierra <alex.sierra@amd.com>
>>>>> ---
>>>>>    arch/powerpc/kvm/book3s_hv_uvmem.c     |  2 +-
>>>>>    drivers/gpu/drm/nouveau/nouveau_dmem.c |  2 +-
>>>>>    fs/dax.c                               |  4 +-
>>>>>    include/linux/dax.h                    |  2 +-
>>>>>    include/linux/memremap.h               |  7 +--
>>>>>    include/linux/mm.h                     | 13 +----
>>>>>    lib/test_hmm.c                         |  2 +-
>>>>>    mm/internal.h                          |  8 +++
>>>>>    mm/memremap.c                          | 68
>>>>> +++++++-------------------
>>>>>    mm/migrate.c                           |  5 --
>>>>>    mm/page_alloc.c                        |  3 ++
>>>>>    mm/swap.c                              | 45 ++---------------
>>>>>    12 files changed, 46 insertions(+), 115 deletions(-)
>>>>>
>>>> I haven't seen a response to the issues I raised back at v3 of this
>>>> series.
>>>> https://lore.kernel.org/linux-mm/4f6dd918-d79b-1aa7-3a4c-caa67ddc29bc@nvidia.com/
>>>>
>>>>
>>>>
>>>> Did I miss something?
>>> I think part of the response was that we did more testing. Alex added
>>> support for DEVICE_GENERIC pages to test_hmm and he ran DAX tests
>>> recommended by Theodore Tso. In that testing he ran into a WARN_ON_ONCE
>>> about a zero page refcount in try_get_page. The fix is in the latest
>>> version of patch 2. But it's already obsolete because John Hubbard is
>>> about to remove that function altogether.
>>>
>>> I think the issues you raised were more uncertainty than known bugs. It
>>> seems the fact that you can have DAX pages with 0 refcount is a feature
>>> more than a bug.
>>>
>>> Regards,
>>>    Felix
>>
>> Did you test on a system without CONFIG_ARCH_HAS_PTE_SPECIAL defined?
>> In that case, mmap() of a DAX device will call insert_page() which calls
>> get_page() which would trigger VM_BUG_ON_PAGE().
>>
>> I can believe it is OK for PTE_SPECIAL page table entries to have no
>> struct page or that MEMORY_DEVICE_GENERIC struct pages be mapped with
>> a zero reference count using insert_pfn().
> Hi Ralph,
> We have tried the DAX tests with and without
> CONFIG_ARCH_HAS_PTE_SPECIAL defined.
> Apparently none of the tests touches that condition for a DAX device.
> Of course,
> that doesn't mean it could happen.
>
> Regards,
> Alex S.
>
>>
>>
>> I find it hard to believe that other MM developers don't see an issue
>> with a struct page with refcount == 0 and mapcount == 1.
>>
>> I don't see where init_page_count() is being called for the
>> MEMORY_DEVICE_GENERIC or MEMORY_DEVICE_PRIVATE struct pages the AMD
>> driver allocates and passes to migrate_vma_setup().
>> Looks like svm_migrate_get_vram_page() needs to call init_page_count()
>> instead of get_page(). (I'm looking at branch
>> origin/alexsierrag/device_generic
>> https://github.com/RadeonOpenCompute/ROCK-Kernel-Driver.git
> Yes, you're right. My bad. Thanks for catching this up. I didn't
> realize I was missing
> to define CONFIG_DEBUG_VM on my build. Therefore this BUG was never
> caught.
> It worked after I replaced get_pages by init_page_count at
> svm_migrate_get_vram_page. However, I don't think this is the best way
> to fix it.
> Ideally, get_pages call should work for device pages with ref count
> equal to 0
> too. Otherwise, we could overwrite refcounter if someone else is
> grabbing the page
> concurrently.

I think using init_page_count in svm_migrate_get_vram_page is the right
answer. This is where the page first gets allocated and initialized
(data migrated into it). I think nobody should have or try to take a
reference to the page before that. We should probably also add a
VM_BUG_ON_PAGE(page_ref_count(page) != 0) before calling init_page_count
to make sure of that.


> I was thinking to add a special condition in get_pages for dev pages.
> This could
> also fix the insert_page -> get_page call from a DAX device.

[+Theodore]

I got lost trying to understand how DAX counts page references and how
the PTE_SPECIAL option affects that. Theodore, can you help with this?
Is there an easy way to test without CONFIG_ARCH_HAS_PTE_SPECIAL on x86,
or do we need to test on a CPU architecture that doesn't support this
feature?

Thanks,
  Felix


>
> Regards,
> Alex S.
>>
>>
>> Also, what about the other places where is_device_private_page() is
>> called?
>> Don't they need to be updated to call is_device_page() instead?
>> One of my goals for this patch was to remove special casing reference
>> counts
>> for ZONE_DEVICE pages in rmap.c, etc.
> Correct, is_device_private_page is still used in rmap, memcontrol and
> migrate.c files
> Looks like rmap and memcontrol should be replaced by is_device_page
> function. However,
> I still need test to validate this. For migrate.c is used in
> remove_migration_pte and
> migrate_vma_insert_page, however these are specific conditions for
> private device type
> Thanks for raise these questions, I think we're getting close.
>
> Regards,
> Alex S.
>>
>> I still think this patch needs an ACK from a FS/DAX maintainer.
>>