Message-ID: <1a406a5f-1336-4051-a722-04b9ea2f54df@nvidia.com>
Date: Thu, 25 Apr 2024 18:23:24 -0700
From: John Hubbard <jhubbard@...dia.com>
To: David Hildenbrand <david@...hat.com>, Matthew Wilcox <willy@...radead.org>
CC: <linux-kernel@...r.kernel.org>, <linux-mm@...ck.org>,
	<linux-doc@...r.kernel.org>, Andrew Morton <akpm@...ux-foundation.org>,
	Jonathan Corbet <corbet@....net>, "Kirill A . Shutemov"
	<kirill.shutemov@...ux.intel.com>, Zi Yan <ziy@...dia.com>, Yang Shi
	<yang.shi@...ux.alibaba.com>, Ryan Roberts <ryan.roberts@....com>
Subject: Re: [PATCH v1] mm/khugepaged: replace page_mapcount() check by
 folio_likely_mapped_shared()

On 4/25/24 1:06 AM, David Hildenbrand wrote:
> On 25.04.24 07:40, John Hubbard wrote:
>> On 4/24/24 9:17 PM, Matthew Wilcox wrote:
>>> On Wed, Apr 24, 2024 at 09:00:50PM -0700, John Hubbard wrote:
..
> We'll talk more about all that at LSF/MM in the mapcount session. A spoiler:

Looking forward to it. And as an aside, this year it feels like the mm
code is changing relatively fast. So many large and small improvements
have happened or are in progress.


> 
> page_mapcount() in the context of large folios:
> * Is a misunderstood function (e.g., page_mapcount() vs page_count()
>    checks, mapped = !page_mapcount() checks).
> * Is a misleading function (e.g., page_mapped() == folio_mapped() but
>    page_mapcount() != folio_mapcount())
> 
> We could just rename it to "folio_precise_page_mapcount()", but then, once we tackle the subpage mapcount optimizations (initially using a separate kernel config toggle), we'll have to teach each caller about an alternative that gets the job done, and it's not that easy to prevent further reuse around the kernel.
> 
> If you look at linux-next, we're down to 5 page_mapcount() calls in fs/proc/, so I'll relocate it to fs/proc/internal.h to prevent any further use - once the s390x change lands in the next merge window.
> 
> Regarding the subpage mapcount optimizations, I can further add:
> * (un)map performance improvements for PTE-mapped THP
> * Preparation for folio_order() > PMD_ORDER, where the current scheme
>    won't scale and needs further adjustments/complexity to even keep it
>    working
> * Preparation for hugetlb-like vmemmap optimizations until we have
>    memdescs / dynamically allocated folios
> * (Paving the way for partially mapping hugetlb folios that faced
>     similar issues? Not sure if that ever gets real, though)
> 
> Is this patch ahead of its time? LSF/MM is just around the corner, and I'm planning on posting the other relevant patches in the next months.

I think so, yes. There is a lot of context required to understand the
motivation, and more required in order to figure out if it is safe,
and if it still provides "good" behavior.

I still think it's mostly harmless, though, so being ahead of its time
is not necessarily an indictment. :)

> 
>>
>>>>> The khugepage MM selftests keep working as expected, including:
>>>>>
>>>>>     Run test: collapse_max_ptes_shared (khugepaged:anon)
>>>>>     Allocate huge page... OK
>>>>>     Share huge page over fork()... OK
>>>>>     Trigger CoW on page 255 of 512... OK
>>>>>     Maybe collapse with max_ptes_shared exceeded.... OK
>>>>>     Trigger CoW on page 256 of 512... OK
>>>>>     Collapse with max_ptes_shared PTEs shared.... OK
>>>>>     Check if parent still has huge page... OK
>>>>
>>>> Well, a word of caution! These tests do not (yet) cover either of
>>>> the interesting new cases that folio_likely_mapped_shared() presents:
>>>> KSM or hugetlbfs interactions. In other words, false positives.
>>>
>>> Hmm ... KSM never uses large folios and hugetlbfs is disjoint from
>>> khugepaged?
> 
> Right, folio_likely_mapped_shared() behaves exactly like page_mapcount() would for (small) KSM folios, so no change for them.
> 
> Thankfully, hugetlb is out of the picture this time.
> 
>>>
>>
>> Oh good. I thought we might have had a testing hole, but no.
> 
> Thanks for having a look!
> 
> I'm only a bit concerned about folio_likely_mapped_shared() "false negatives" (detecting exclusive although shared), until we have a more precise folio_likely_mapped_shared() variant to not unexpectedly waste memory.
> 
> Imagine someone would be setting "khugepaged_max_ptes_shared=0", and then we have an area where (I think this is the extreme case):
> 
> * We map 256 subpages of a 2M folio that are shared 256 times with a
>    child process.
> * We don't map the first subpage.
> * One PTE maps another page and is pte_write().
> * 255 PTEs are pte_none().
> 
> folio_likely_mapped_shared() would return "false".
> 
> But then my thinking is:
> * We are already wasting 256 subpages that are free in the 2M folio.
>    Sure, we might be able to relaim it when deferred splitting.
> * Why would someone set khugepaged_max_ptes_shared=0 but leave
>    khugepaged_max_ptes_none set that high that we would allow 255
>    pte_none?
> * If the child is a short-living subprocess, we don't really care
> * Any futher writes to unbacked/R/O PTEs in that PMD area would COW and
>    consume memory.
> 
> So I had to use more and more "ifs" to construct a scenario where we might end up wasting 1M of memory, at which point I decided "this is really a corner case" and likely not worth the worry.
> 
> If we run into real issues, though, it will be easy to just inline page_mapcount() here to resolve it; but the less special-casing the better.
> 

OK. I'll need to think through some more of these cases. And meanwhile, I
was poking around from the other direction: just injection test it by
pasting in "true" or "false", in place of calling folio_likely_mapped_shared().
And see what changes.

The "true" test lets me fail several khugepaged selftests, while the "false"
test just increases the counter in /proc/vmstat.

That's more of a black box way of poking at it, just to have another facet
of testing. Because it is good to ensure that we really do have test
coverage if we're changing the code. Anyway, just ideas.


thanks,
-- 
John Hubbard
NVIDIA

Powered by blists - more mailing lists