All of lore.kernel.org
 help / color / mirror / Atom feed
From: Alex Thorlton <athorlton@sgi.com>
To: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@redhat.com>, Andi Kleen <andi@firstfloor.org>,
	Alex Thorlton <athorlton@sgi.com>,
	linux-mm@kvack.org, Andrew Morton <akpm@linux-foundation.org>,
	Bob Liu <lliubbo@gmail.com>, David Rientjes <rientjes@google.com>,
	"Eric W. Biederman" <ebiederm@xmission.com>,
	Hugh Dickins <hughd@google.com>, Ingo Molnar <mingo@redhat.com>,
	Kees Cook <keescook@chromium.org>,
	"Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
	Mel Gorman <mgorman@suse.de>, Oleg Nesterov <oleg@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Vladimir Davydov <vdavydov@parallels.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH 0/4] Convert khugepaged to a task_work function
Date: Mon, 17 Nov 2014 15:34:15 -0600	[thread overview]
Message-ID: <20141117213415.GU21147@sgi.com> (raw)
In-Reply-To: <5453F0A4.4090708@suse.cz>

On Fri, Oct 31, 2014 at 09:27:16PM +0100, Vlastimil Babka wrote:
> What could help would be to cache one or few free huge pages per
> zone with cache
> re-fill done asynchronously, e.g. via work queues. The cache could
> benefit fault-THP
> allocations as well. And adding some logic that if nobody uses the
> cached pages and
> memory is low, then free them. And importantly, if it's not possible
> to allocate huge
> pages for the cache, then prevent scanning for collapse candidates
> as there's no point.
> (well this is probably more complex if some nodes can allocate huge
> pages and others
> not).

I think this would be a pretty cool addition, even separately from this
effort.  If we keep a page cached on each NUMA node, then we could,
theoretically, really speed up the khugepaged scans (even if we don't
move those scans to task_work), and regular THP faults.  I'll add it to
my ever-growing wish list :)

- Alex

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

WARNING: multiple messages have this Message-ID (diff)
From: Alex Thorlton <athorlton@sgi.com>
To: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@redhat.com>, Andi Kleen <andi@firstfloor.org>,
	Alex Thorlton <athorlton@sgi.com>,
	linux-mm@kvack.org, Andrew Morton <akpm@linux-foundation.org>,
	Bob Liu <lliubbo@gmail.com>, David Rientjes <rientjes@google.com>,
	"Eric W. Biederman" <ebiederm@xmission.com>,
	Hugh Dickins <hughd@google.com>, Ingo Molnar <mingo@redhat.com>,
	Kees Cook <keescook@chromium.org>,
	"Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
	Mel Gorman <mgorman@suse.de>, Oleg Nesterov <oleg@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Vladimir Davydov <vdavydov@parallels.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH 0/4] Convert khugepaged to a task_work function
Date: Mon, 17 Nov 2014 15:34:15 -0600	[thread overview]
Message-ID: <20141117213415.GU21147@sgi.com> (raw)
In-Reply-To: <5453F0A4.4090708@suse.cz>

On Fri, Oct 31, 2014 at 09:27:16PM +0100, Vlastimil Babka wrote:
> What could help would be to cache one or few free huge pages per
> zone with cache
> re-fill done asynchronously, e.g. via work queues. The cache could
> benefit fault-THP
> allocations as well. And adding some logic that if nobody uses the
> cached pages and
> memory is low, then free them. And importantly, if it's not possible
> to allocate huge
> pages for the cache, then prevent scanning for collapse candidates
> as there's no point.
> (well this is probably more complex if some nodes can allocate huge
> pages and others
> not).

I think this would be a pretty cool addition, even separately from this
effort.  If we keep a page cached on each NUMA node, then we could,
theoretically, really speed up the khugepaged scans (even if we don't
move those scans to task_work), and regular THP faults.  I'll add it to
my ever-growing wish list :)

- Alex

  reply	other threads:[~2014-11-17 21:33 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-10-23  2:49 [PATCH 0/4] Convert khugepaged to a task_work function Alex Thorlton
2014-10-23  2:49 ` Alex Thorlton
2014-10-23  2:49 ` [PATCH 1/4] Disable khugepaged thread Alex Thorlton
2014-10-23  2:49   ` Alex Thorlton
2014-10-23  2:49 ` [PATCH] Add pgcollapse controls to task_struct Alex Thorlton
2014-10-23  2:49   ` Alex Thorlton
2014-10-23 15:29   ` Alex Thorlton
2014-10-23 15:29     ` Alex Thorlton
2014-10-23  2:49 ` [PATCH 3/4] Convert khugepaged scan functions to work with task_work Alex Thorlton
2014-10-23  2:49 ` [PATCH 4/4] Add /proc files to expose per-mm pgcollapse stats Alex Thorlton
2014-10-23  2:49   ` Alex Thorlton
2014-10-23  3:06 ` [PATCH 1/2] Add pgcollapse stat counter to task_struct Alex Thorlton
2014-10-23  3:06   ` Alex Thorlton
2014-10-23  3:06 ` [PATCH 2/2] Add /proc files to expose per-mm pgcollapse stats Alex Thorlton
2014-10-23  3:06   ` Alex Thorlton
2014-10-23 17:55 ` [PATCH 0/4] Convert khugepaged to a task_work function Rik van Riel
2014-10-23 17:55   ` Rik van Riel
2014-10-23 18:05   ` Alex Thorlton
2014-10-23 18:05     ` Alex Thorlton
2014-10-23 18:52     ` Alex Thorlton
2014-10-23 18:52       ` Alex Thorlton
2014-10-28 12:12 ` Andi Kleen
2014-10-28 12:12   ` Andi Kleen
2014-10-28 12:58   ` Rik van Riel
2014-10-28 12:58     ` Rik van Riel
2014-10-28 15:39     ` Rik van Riel
2014-10-28 15:39       ` Rik van Riel
2014-10-31 20:27       ` Vlastimil Babka
2014-10-31 20:27         ` Vlastimil Babka
2014-11-17 21:34         ` Alex Thorlton [this message]
2014-11-17 21:34           ` Alex Thorlton
2014-11-10 11:03     ` Mel Gorman
2014-11-10 11:03       ` Mel Gorman
2014-11-17 21:16       ` Alex Thorlton
2014-11-17 21:16         ` Alex Thorlton
2014-10-29 21:58   ` Alex Thorlton
2014-10-29 21:58     ` Alex Thorlton
2014-10-30  0:23     ` Kirill A. Shutemov
2014-10-30  0:23       ` Kirill A. Shutemov
2014-10-30  8:35     ` Andi Kleen
2014-10-30  8:35       ` Andi Kleen
2014-10-30 18:25       ` Alex Thorlton
2014-10-30 18:25         ` Alex Thorlton
  -- strict thread matches above, loose matches on Subject: below --
2014-10-23  2:35 Alex Thorlton
2014-10-23  2:43 ` Alex Thorlton

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=20141117213415.GU21147@sgi.com \
    --to=athorlton@sgi.com \
    --cc=akpm@linux-foundation.org \
    --cc=andi@firstfloor.org \
    --cc=ebiederm@xmission.com \
    --cc=hughd@google.com \
    --cc=keescook@chromium.org \
    --cc=kirill.shutemov@linux.intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=lliubbo@gmail.com \
    --cc=mgorman@suse.de \
    --cc=mingo@redhat.com \
    --cc=oleg@redhat.com \
    --cc=peterz@infradead.org \
    --cc=riel@redhat.com \
    --cc=rientjes@google.com \
    --cc=tglx@linutronix.de \
    --cc=vbabka@suse.cz \
    --cc=vdavydov@parallels.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 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.