All of lore.kernel.org
 help / color / mirror / Atom feed
From: Hao Peng <flyingpenghao@gmail.com>
To: David Hildenbrand <david@redhat.com>
Cc: akpm@linux-foundation.org, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mm/huge_memory: disable thp if thp page size is too large
Date: Wed, 6 Oct 2021 16:16:51 +0800	[thread overview]
Message-ID: <CAPm50aKRTg2=WMDw+vWvXA5+wADfSo8Kk-PHpAagftGypArU+g@mail.gmail.com> (raw)
In-Reply-To: <914a0b38-cb44-f4be-43aa-bc42bee1553b@redhat.com>

On Thu, Sep 30, 2021 at 9:19 PM David Hildenbrand <david@redhat.com> wrote:
>
> On 30.09.21 04:14, Hao Peng wrote:
> > From: Peng Hao <flyingpeng@tencent.com>
>
> "disable thp if thp page size is too large" you disable thp if there is
> not sufficient memory installed in the system (and for that, you use THP
> size), something doesn't add up here.
>
In addition, if THP is enabled, min_free_kbytes will be recalculated.
At this time,
min_free_kbytes is calculated based on the size of THP. If the size of
THP is too l
arge, OOM will easily be triggered.
> >
> > After seting the page size to 64k on ARM64, the supported huge page
>
> s/seting/setting/
>
> Fortunately, most distributions already switched to 4k, because 512MB
> THP is pretty much useless, especially on any system that doesn't have
> memory in the range of hundreds of megabytes or terrabytes.
But the ARM64 Server distribution version I use, such as CentOS for ARM64,
 has a page size of 16KB or 64KB.
Thanks.
>
> > size is 512M and 1TB. Therefore, if the thp is enabled, the size
> > of the thp is 512M. In this case, min_free_kbytes will be too large.
> >
> > On an arm64 server with 64G memory, the page size is 64k, with thp
> > enabled.
> > cat /proc/sys/vm/min_free_kbytes
> > 3335104
> >
> > Therefore, when judging whether to enable thp by default, consider
> > the size of thp.
> >
> > Signed-off-by: Peng Hao <flyingpeng@tencent.com>
> > ---
> >   mm/huge_memory.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> > index 5e9ef0fc261e..03c7f571b3ae 100644
> > --- a/mm/huge_memory.c
> > +++ b/mm/huge_memory.c
> > @@ -437,7 +437,7 @@ static int __init hugepage_init(void)
> >           * where the extra memory used could hurt more than TLB overhead
> >           * is likely to save.  The admin can still enable it through /sys.
> >           */
> > -       if (totalram_pages() < (512 << (20 - PAGE_SHIFT))) {
> > +       if (totalram_pages() < (512 << (HPAGE_PMD_SHIFT - PAGE_SHIFT))) {
> >                  transparent_hugepage_flags = 0;
> >                  return 0;
> >          }
> > --
> > 2.27.0
> >
>
>
> --
> Thanks,
>
> David / dhildenb
>

  reply	other threads:[~2021-10-06  8:17 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-09-30  2:14 [PATCH] mm/huge_memory: disable thp if thp page size is too large Hao Peng
2021-09-30 13:19 ` David Hildenbrand
2021-10-06  8:16   ` Hao Peng [this message]
2021-10-06  8:20     ` David Hildenbrand

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='CAPm50aKRTg2=WMDw+vWvXA5+wADfSo8Kk-PHpAagftGypArU+g@mail.gmail.com' \
    --to=flyingpenghao@gmail.com \
    --cc=akpm@linux-foundation.org \
    --cc=david@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.