From: "Huang\, Ying" <ying.huang@intel.com>
To: David Hildenbrand <david@redhat.com>
Cc: Michal Hocko <mhocko@kernel.org>,
Matthew Wilcox <willy@infradead.org>,
Andrew Morton <akpm@linux-foundation.org>, <linux-mm@kvack.org>,
<linux-kernel@vger.kernel.org>, Mel Gorman <mgorman@suse.de>,
Vlastimil Babka <vbabka@suse.cz>, Zi Yan <ziy@nvidia.com>,
Peter Zijlstra <peterz@infradead.org>,
Dave Hansen <dave.hansen@linux.intel.com>,
"Minchan Kim" <minchan@kernel.org>,
Johannes Weiner <hannes@cmpxchg.org>,
Hugh Dickins <hughd@google.com>,
Alexander Duyck <alexander.duyck@gmail.com>
Subject: Re: [RFC 0/3] mm: Discard lazily freed pages when migrating
Date: Tue, 03 Mar 2020 08:25:43 +0800 [thread overview]
Message-ID: <878ski708o.fsf@yhuang-dev.intel.com> (raw)
In-Reply-To: <8005e5a1-e2f2-1e57-ccb4-0cb9371b080d@redhat.com> (David Hildenbrand's message of "Mon, 2 Mar 2020 15:23:16 +0100")
David Hildenbrand <david@redhat.com> writes:
> On 02.03.20 15:12, Huang, Ying wrote:
>> Michal Hocko <mhocko@kernel.org> writes:
>>
>>> On Fri 28-02-20 16:55:40, Huang, Ying wrote:
>>>> David Hildenbrand <david@redhat.com> writes:
>>> [...]
>>>>> E.g., free page reporting in QEMU wants to use MADV_FREE. The guest will
>>>>> report currently free pages to the hypervisor, which will MADV_FREE the
>>>>> reported memory. As long as there is no memory pressure, there is no
>>>>> need to actually free the pages. Once the guest reuses such a page, it
>>>>> could happen that there is still the old page and pulling in in a fresh
>>>>> (zeroed) page can be avoided.
>>>>>
>>>>> AFAIKs, after your change, we would get more pages discarded from our
>>>>> guest, resulting in more fresh (zeroed) pages having to be pulled in
>>>>> when a guest touches a reported free page again. But OTOH, page
>>>>> migration is speed up (avoiding to migrate these pages).
>>>>
>>>> Let's look at this problem in another perspective. To migrate the
>>>> MADV_FREE pages of the QEMU process from the node A to the node B, we
>>>> need to free the original pages in the node A, and (maybe) allocate the
>>>> same number of pages in the node B. So the question becomes
>>>>
>>>> - we may need to allocate some pages in the node B
>>>> - these pages may be accessed by the application or not
>>>> - we should allocate all these pages in advance or allocate them lazily
>>>> when they are accessed.
>>>>
>>>> We thought the common philosophy in Linux kernel is to allocate lazily.
>>>
>>> The common philosophy is to cache as much as possible.
>>
>> Yes. This is another common philosophy. And MADV_FREE pages is
>> different from caches such as the page caches because it has no valid
>> contents.
>
> Side note: It might contain valid content until discarded/zeroed out.
> E.g., an application could use a marker bit (e.g., first bit) to detect
> if the page still contains valid data or not. If the data is still
> marked valid, the content could be reuse immediately. Not sure if there
> is any such application, though :)
I don't think this is the typical use case. But I admit that this is
possible.
Best Regards,
Huang, Ying
next prev parent reply other threads:[~2020-03-03 0:25 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-02-28 3:38 [RFC 0/3] mm: Discard lazily freed pages when migrating Huang, Ying
2020-02-28 3:38 ` [RFC 1/3] mm, migrate: Check return value of try_to_unmap() Huang, Ying
2020-02-28 3:38 ` [RFC 2/3] mm: Add a new page flag PageLayzyFree() for MADV_FREE Huang, Ying
2020-02-28 6:13 ` David Hildenbrand
2020-02-28 6:47 ` Huang, Ying
2020-03-15 8:18 ` Wei Yang
2020-03-15 8:54 ` Mika Penttilä
2020-03-15 12:22 ` Wei Yang
2020-03-16 1:21 ` Huang, Ying
2020-03-16 22:38 ` Wei Yang
2020-02-28 3:38 ` [RFC 3/3] mm: Discard lazily freed pages when migrating Huang, Ying
2020-02-28 3:42 ` [RFC 0/3] " Matthew Wilcox
2020-02-28 7:25 ` Huang, Ying
2020-02-28 8:22 ` David Hildenbrand
2020-02-28 8:55 ` Huang, Ying
2020-02-28 9:49 ` Mel Gorman
2020-03-02 11:23 ` Huang, Ying
2020-03-02 15:16 ` Mel Gorman
2020-03-03 1:51 ` Huang, Ying
2020-03-03 8:09 ` Michal Hocko
2020-03-03 8:47 ` Huang, Ying
2020-03-03 8:58 ` Michal Hocko
2020-03-03 11:49 ` Huang, Ying
2020-03-04 9:58 ` Michal Hocko
2020-03-04 10:56 ` Mel Gorman
2020-03-05 1:42 ` Huang, Ying
2020-03-04 11:15 ` Huang, Ying
2020-03-04 11:26 ` Michal Hocko
2020-03-05 1:45 ` Huang, Ying
2020-03-05 10:48 ` Mel Gorman
2020-03-06 4:05 ` Huang, Ying
2020-03-09 5:26 ` Huang, Ying
2020-03-03 13:02 ` Mel Gorman
2020-03-04 0:33 ` Huang, Ying
2020-02-28 9:50 ` Michal Hocko
2020-02-28 10:15 ` Michal Hocko
2020-02-28 13:45 ` Johannes Weiner
2020-03-02 14:12 ` Huang, Ying
2020-03-02 14:23 ` David Hildenbrand
2020-03-03 0:25 ` Huang, Ying [this message]
2020-03-02 14:25 ` Michal Hocko
2020-03-03 1:30 ` Huang, Ying
2020-03-03 8:19 ` Michal Hocko
2020-03-03 11:36 ` Huang, Ying
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=878ski708o.fsf@yhuang-dev.intel.com \
--to=ying.huang@intel.com \
--cc=akpm@linux-foundation.org \
--cc=alexander.duyck@gmail.com \
--cc=dave.hansen@linux.intel.com \
--cc=david@redhat.com \
--cc=hannes@cmpxchg.org \
--cc=hughd@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mgorman@suse.de \
--cc=mhocko@kernel.org \
--cc=minchan@kernel.org \
--cc=peterz@infradead.org \
--cc=vbabka@suse.cz \
--cc=willy@infradead.org \
--cc=ziy@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).