linux-kernel - Re: [LKP] Re: [mm/gup] 57efa1fe59: will-it-scale.per_thread

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Fri, 4 Jun 2021 16:37:53 +0800
From:   Xing Zhengjun <zhengjun.xing@...ux.intel.com>
To:     Linus Torvalds <torvalds@...ux-foundation.org>,
        kernel test robot <oliver.sang@...el.com>
Cc:     Jason Gunthorpe <jgg@...dia.com>,
        John Hubbard <jhubbard@...dia.com>, Jan Kara <jack@...e.cz>,
        Peter Xu <peterx@...hat.com>,
        Andrea Arcangeli <aarcange@...hat.com>,
        "Aneesh Kumar K.V" <aneesh.kumar@...ux.ibm.com>,
        Christoph Hellwig <hch@....de>,
        Hugh Dickins <hughd@...gle.com>, Jann Horn <jannh@...gle.com>,
        Kirill Shutemov <kirill@...temov.name>,
        Kirill Tkhai <ktkhai@...tuozzo.com>,
        Leon Romanovsky <leonro@...dia.com>,
        Michal Hocko <mhocko@...e.com>,
        Oleg Nesterov <oleg@...hat.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        LKML <linux-kernel@...r.kernel.org>, lkp@...ts.01.org,
        kernel test robot <lkp@...el.com>
Subject: Re: [LKP] Re: [mm/gup] 57efa1fe59: will-it-scale.per_thread_ops -9.2%
 regression

Hi Linus,

On 5/25/2021 11:11 AM, Linus Torvalds wrote:
> On Mon, May 24, 2021 at 5:00 PM kernel test robot <oliver.sang@...el.com> wrote:
>> FYI, we noticed a -9.2% regression of will-it-scale.per_thread_ops due to commit:
>> commit: 57efa1fe5957694fa541c9062de0a127f0b9acb0 ("mm/gup: prevent gup_fast from racing with COW during fork")
> Hmm. This looks like one of those "random fluctuations" things.
>
> It would be good to hear if other test-cases also bisect to the same
> thing, but this report already says:
>
>> In addition to that, the commit also has significant impact on the following tests:
>>
>> +------------------+---------------------------------------------------------------------------------+
>> | testcase: change | will-it-scale: will-it-scale.per_thread_ops 3.7% improvement                    |
> which does kind of reinforce that "this benchmark gives unstable numbers".
>
> The perf data doesn't even mention any of the GUP paths, and on the
> pure fork path the biggest impact would be:
>
>   (a) maybe "struct mm_struct" changed in size or had a different cache layout
I move "write_protect_seq" to the tail of the "struct mm_struct", the 
regression reduced to -3.6%. The regression should relate to the cache 
layout.
=========================================================================================
tbox_group/testcase/rootfs/kconfig/compiler/nr_task/mode/test/cpufreq_governor/ucode:
lkp-icl-2sp1/will-it-scale/debian-10.4-x86_64-20200603.cgz/x86_64-rhel-8.3/gcc-9/50%/thread/mmap1/performance/0xb000280

commit:
   c28b1fc70390df32e29991eedd52bd86e7aba080
   57efa1fe5957694fa541c9062de0a127f0b9acb0
   f6a9c27882d51ff551e15522992d3725c342372d  (the test patch)

c28b1fc70390df32 57efa1fe5957694fa541c9062de f6a9c27882d51ff551e15522992
---------------- --------------------------- ---------------------------
          %stddev     %change         %stddev     %change %stddev
              \          |                \          | \
     341938            -9.0%     311218 ±  2%      -3.6% 329513        
will-it-scale.48.threads
       7123            -9.0%       6483 ±  2%      -3.6% 6864        
will-it-scale.per_thread_ops
     341938            -9.0%     311218 ±  2%      -3.6% 329513        
will-it-scale.workload

diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index 915f4f100383..34bb2a01806c 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -447,13 +447,6 @@ struct mm_struct {
                  */
                 atomic_t has_pinned;

-               /**
-                * @write_protect_seq: Locked when any thread is write
-                * protecting pages mapped by this mm to enforce a later 
COW,
-                * for instance during page table copying for fork().
-                */
-               seqcount_t write_protect_seq;
-
  #ifdef CONFIG_MMU
                 atomic_long_t pgtables_bytes;   /* PTE page table pages */
  #endif
@@ -564,6 +557,12 @@ struct mm_struct {
  #ifdef CONFIG_IOMMU_SUPPORT
                 u32 pasid;
  #endif
+                /**
+                 * @write_protect_seq: Locked when any thread is write
+                 * protecting pages mapped by this mm to enforce a 
later COW,
+                 * for instance during page table copying for fork().
+                 */
+                seqcount_t write_protect_seq;
         } __randomize_layout;

         /*

>
>   (b) two added (nonatomic) increment operations in the fork path due
> to the seqcount
>
> and I'm not seeing what would cause that 9% change. Obviously cache
> placement has done it before.
>
> If somebody else sees something that I'm missing, please holler. But
> I'll ignore this as "noise" otherwise.
>
>              Linus
> _______________________________________________
> LKP mailing list -- lkp@...ts.01.org
> To unsubscribe send an email to lkp-leave@...ts.01.org

-- 
Zhengjun Xing