Re: [PATCH v2] proc/ksm: add ksm stats to /proc/pid/smaps

[Stefan Roesch <shr@devkernel.io>]

David Hildenbrand <david@redhat.com> writes:

> On 15.08.23 19:10, Stefan Roesch wrote:
>> David Hildenbrand <david@redhat.com> writes:
>>
>>> Sorry for the late reply, Gmail once again decided to classify your mails as
>>> spam (for whatever reason).
>>>
>>> On 11.08.23 18:28, Stefan Roesch wrote:
>>>> With madvise and prctl KSM can be enabled for different VMA's. Once it
>>>> is enabled we can query how effective KSM is overall. However we cannot
>>>> easily query if an individual VMA benefits from KSM.
>>>> This commit adds a KSM section to the /prod/<pid>/smaps file. It reports
>>>> how many of the pages are KSM pages.
>>>> Here is a typical output:
>>>> 7f420a000000-7f421a000000 rw-p 00000000 00:00 0
>>>> Size:             262144 kB
>>>> KernelPageSize:        4 kB
>>>> MMUPageSize:           4 kB
>>>> Rss:               51212 kB
>>>> Pss:                8276 kB
>>>> Shared_Clean:        172 kB
>>>> Shared_Dirty:      42996 kB
>>>> Private_Clean:       196 kB
>>>> Private_Dirty:      7848 kB
>>>> Referenced:        15388 kB
>>>> Anonymous:         51212 kB
>>>> KSM:               41376 kB
>>>> LazyFree:              0 kB
>>>> AnonHugePages:         0 kB
>>>> ShmemPmdMapped:        0 kB
>>>> FilePmdMapped:         0 kB
>>>> Shared_Hugetlb:        0 kB
>>>> Private_Hugetlb:       0 kB
>>>> Swap:             202016 kB
>>>> SwapPss:            3882 kB
>>>> Locked:                0 kB
>>>> THPeligible:    0
>>>> ProtectionKey:         0
>>>> ksm_state:          0
>>>> ksm_skip_base:      0
>>>> ksm_skip_count:     0
>>>> VmFlags: rd wr mr mw me nr mg anon
>>>> This information also helps with the following workflow:
>>>> - First enable KSM for all the VMA's of a process with prctl.
>>>> - Then analyze with the above smaps report which VMA's benefit the most
>>>> - Change the application (if possible) to add the corresponding madvise
>>>> calls for the VMA's that benefit the most
>>>> Signed-off-by: Stefan Roesch <shr@devkernel.io>
>>>> ---
>>>>    Documentation/filesystems/proc.rst | 3 +++
>>>>    fs/proc/task_mmu.c                 | 5 +++++
>>>>    2 files changed, 8 insertions(+)
>>>> diff --git a/Documentation/filesystems/proc.rst
>>>> b/Documentation/filesystems/proc.rst
>>>> index 7897a7dafcbc..4ef3c0bbf16a 100644
>>>> --- a/Documentation/filesystems/proc.rst
>>>> +++ b/Documentation/filesystems/proc.rst
>>>> @@ -461,6 +461,7 @@ Memory Area, or VMA) there is a series of lines such as the following::
>>>>        Private_Dirty:         0 kB
>>>>        Referenced:          892 kB
>>>>        Anonymous:             0 kB
>>>> +    KSM:                   0 kB
>>>>        LazyFree:              0 kB
>>>>        AnonHugePages:         0 kB
>>>>        ShmemPmdMapped:        0 kB
>>>> @@ -501,6 +502,8 @@ accessed.
>>>>    a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE
>>>>    and a page is modified, the file page is replaced by a private anonymous copy.
>>>>    +"KSM" shows the amount of anonymous memory that has been de-duplicated.
>>>
>>>
>>> How do we want to treat memory that has been deduplicated into the shared
>>> zeropage?
>>>
>>> It would also match this description.
>>>
>>> See in mm-stable:
>>>
>>> commit 30ff6ed9a65c7e73545319fc15f7bcf9c52457eb
>>> Author: xu xin <xu.xin16@zte.com.cn>
>>> Date:   Tue Jun 13 11:09:28 2023 +0800
>>>
>>>      ksm: support unsharing KSM-placed zero pages
>>>
>>>      Patch series "ksm: support tracking KSM-placed zero-pages", v10.
>> I see two approaches how to deal with zero page:
>>   - If zero page is not enabled, it works as is
>>   - If enabled
>>      - Document that zero page is accounted for the current vma or
>>      - Pass in the pte from smaps_pte_entry() to smaps_account() so we can
>>      determine if this is a zero page.
>
> That's probably the right thing to do: make the stat return the same value
> independent of the usage of the shared zeropage.
>

I'll update the documentation accordingly.

>>      I'm not sure what to do about smaps_pmd_entry in that case. We
>>      probably don't care about compund pages.
>
> No, KSM only places the shared zeropage for PTEs, no need to handle PMDs.