linux-f2fs-devel.lists.sourceforge.net archive mirror
 help / color / mirror / Atom feed
From: Chao Yu <chao2.yu@samsung.com>
To: Jaegeuk Kim <jaegeuk@kernel.org>, Changman Lee <cm224.lee@samsung.com>
Cc: linux-kernel@vger.kernel.org, linux-f2fs-devel@lists.sourceforge.net
Subject: [PATCH 0/5] f2fs: support inline dir
Date: Sat, 09 Aug 2014 10:44:31 +0800	[thread overview]
Message-ID: <00be01cfb37b$f3ba83f0$db2f8bd0$@samsung.com> (raw)

There are large space in f2fs inode, so last inline data patch set have made
f2fs to acquire the ability of storing data of small file (less than ~3.4k)
directly in inode block, but not support inline dir. In this patch, we make
f2fs to support inline dir.

Layout:
Inline dir layout is the same as inline data layout,
a. i_addr[0] are reserved space for converting from inline dir to regular one
when out-of-space occur in inline dentry.
b. i_addr[1..872] will be used as storing space of inline dentry.
c. i_addr[873..922] are reserved for inline xattr.

Size:
Since our max size of inline dir space is limit to MAX_INLINE_DATA (3488 bytes), 
we introduce inline dentry struct fit for available space, then our dentry slot
number in inline dentry reduce to 182, less than normal dentry block's 214.

Process:
In inline dir mode, our dir entries will be lookuped/stored/deleted in the
inline dentry space of inode, util there are no more space to store new added
dir entry, in this case we will convert inline dentry to normal 0-index dentry
block and disable inline dir mode for this inode.

Benefit points:
a) space saving
   1) Test with fsstress with special arguments, it can save about 3% space.
   time fsstress -c -p 20 -n 500 -l 10 -d /mnt/f2fs -w -f chown=0 -f creat=10
   -f dwrite=0 -f fdatasync=0 -f fsync=0 -f link=10 -f mkdir=10 -f mknod=10
   -f rename=0 -f rmdir=0 -f symlink=10 -f truncate=0 -f unlink=0 -f write=10 -S
   based	/dev/sdb        20969472  2234136  18591976  11% /mnt/f2fs
   patched	/dev/sdb        20969472  2303120  18522992  12% /mnt/f2fs
   2) Test with storing kernel src, it can save less than 1% space.

b) performance
   Test with fsstress shows cost time reduce 21%.
   time fsstress -d /mnt/f2fs -l 5 -n 1000 -p 20 -c -r
   based	168.503 s  (inline data)
   patched	132.840 s  (inline data + inline dir)

Chao Yu (5):
  f2fs: add infra macro and sturct for inline dir
  f2fs: export dir operations for inline dir
  f2fs: add key function to handle inline dir
  f2fs: enable inline dir handling
  f2fs: update f2fs documentation

 Documentation/filesystems/f2fs.txt |   5 +-
 fs/f2fs/dir.c                      |  48 +++--
 fs/f2fs/f2fs.h                     |  18 +-
 fs/f2fs/inline.c                   | 388 +++++++++++++++++++++++++++++++++++++
 fs/f2fs/namei.c                    |   8 +-
 fs/f2fs/recovery.c                 |   2 +-
 include/linux/f2fs_fs.h            |  18 ++
 7 files changed, 469 insertions(+), 18 deletions(-)

-- 
2.0.1.474.g72c7794



------------------------------------------------------------------------------

                 reply	other threads:[~2014-08-09  2:45 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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='00be01cfb37b$f3ba83f0$db2f8bd0$@samsung.com' \
    --to=chao2.yu@samsung.com \
    --cc=cm224.lee@samsung.com \
    --cc=jaegeuk@kernel.org \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).