linux-kernel - Re: [PATCH v3 2/2] tmpfs: Make tmpfs scalable with percpu

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: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <AANLkTikOpTsdxhjqE0G3oEv9qT2nRFp3y5fjDUqXlw4X@mail.gmail.com>
Date:	Fri, 18 Jun 2010 10:35:51 +0900
From:	Minchan Kim <minchan.kim@...il.com>
To:	Tim Chen <tim.c.chen@...ux.intel.com>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	linux-kernel@...r.kernel.org, Andi Kleen <ak@...ux.intel.com>,
	Hugh Dickins <hughd@...gle.com>, yanmin.zhang@...el.com
Subject: Re: [PATCH v3 2/2] tmpfs: Make tmpfs scalable with percpu_counter for 
	used blocks

Hello.
The idea is good. :)

On Fri, Jun 18, 2010 at 8:56 AM, Tim Chen <tim.c.chen@...ux.intel.com> wrote:
> The current implementation of tmpfs is not scalable.
> We found that stat_lock is contended by multiple threads
> when we need to get a new page, leading to useless spinning
> inside this spin lock.
>
> This patch makes use of the percpu_counter library to maintain local
> count of used blocks to speed up getting and returning
> of pages.  So the acquisition of stat_lock is unnecessary
> for getting and returning blocks, improving the performance
> of tmpfs on system with large number of cpus.  On a 4 socket
> 32 core NHM-EX system, we saw improvement of 270%.
>

Good enhancement. :)

> Signed-off-by: Tim Chen <tim.c.chen@...ux.intel.com>
>  include/linux/shmem_fs.h |    3 ++-
>  mm/shmem.c               |   40 +++++++++++++++++-----------------------
>  2 files changed, 19 insertions(+), 24 deletions(-)
>
> diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
> index e164291..399be5a 100644
> --- a/include/linux/shmem_fs.h
> +++ b/include/linux/shmem_fs.h
> @@ -3,6 +3,7 @@
>
>  #include <linux/swap.h>
>  #include <linux/mempolicy.h>
> +#include <linux/percpu_counter.h>
>
>  /* inode in-kernel data */
>
> @@ -23,7 +24,7 @@ struct shmem_inode_info {
>
>  struct shmem_sb_info {
>        unsigned long max_blocks;   /* How many blocks are allowed */
> -       unsigned long free_blocks;  /* How many are left for allocation */
> +       struct percpu_counter used_blocks;  /* How many are allocated */

Just a nitpick.
Why do you change free_blocks and used_blocks?
I think we can use free_blocks following as.

ex)
if (percpu_counter_compare(&sbinfo->free_blocks, 0))

Do you have any reason?
Please, justify it.

Thanks, Tim.
-- 
Kind regards,
Minchan Kim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/