lkml.org 
[lkml]   [2012]   [Feb]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    From
    Date
    SubjectRe: [PATCH 6/6 v4] memcg: fix performance of mem_cgroup_begin_update_page_stat()
    On Mon, Feb 13, 2012 at 7:16 PM, KAMEZAWA Hiroyuki
    <kamezawa.hiroyu@jp.fujitsu.com> wrote:
    > From 3377fd7b6e23a5d2a368c078eae27e2b49c4f4aa Mon Sep 17 00:00:00 2001
    > From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
    > Date: Mon, 6 Feb 2012 12:14:47 +0900
    > Subject: [PATCH 6/6] memcg: fix performance of mem_cgroup_begin_update_page_stat()
    >
    > mem_cgroup_begin_update_page_stat() should be very fast because
    > it's called very frequently. Now, it needs to look up page_cgroup
    > and its memcg....this is slow.
    >
    > This patch adds a global variable to check "a memcg is moving or not".

    s/a memcg/any memcg/

    > By this, the caller doesn't need to visit page_cgroup and memcg.

    s/By/With/

    > Here is a test result. A test program makes page faults onto a file,
    > MAP_SHARED and makes each page's page_mapcount(page) > 1, and free
    > the range by madvise() and page fault again.  This program causes
    > 26214400 times of page fault onto a file(size was 1G.) and shows
    > shows the cost of mem_cgroup_begin_update_page_stat().

    Out of curiosity, what is the performance of the mmap program before
    this series?

    > Before this patch for mem_cgroup_begin_update_page_stat()
    > [kamezawa@bluextal test]$ time ./mmap 1G
    >
    > real    0m21.765s
    > user    0m5.999s
    > sys     0m15.434s
    >
    >    27.46%     mmap  mmap               [.] reader
    >    21.15%     mmap  [kernel.kallsyms]  [k] page_fault
    >     9.17%     mmap  [kernel.kallsyms]  [k] filemap_fault
    >     2.96%     mmap  [kernel.kallsyms]  [k] __do_fault
    >     2.83%     mmap  [kernel.kallsyms]  [k] __mem_cgroup_begin_update_page_stat
    >
    > After this patch
    > [root@bluextal test]# time ./mmap 1G
    >
    > real    0m21.373s
    > user    0m6.113s
    > sys     0m15.016s
    >
    > In usual path, calls to __mem_cgroup_begin_update_page_stat() goes away.
    >
    > Note: we may be able to remove this optimization in future if
    >      we can get pointer to memcg directly from struct page.
    >
    > Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

    Acked-by: Greg Thelen <gthelen@google.com>
    --
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

    \
     
     \ /
      Last update: 2012-02-14 08:25    [W:0.024 / U:121.276 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site