public inbox for linux-btrfs@vger.kernel.org
 help / color / mirror / Atom feed
From: Wang Yugui <wangyugui@e16-tech.com>
To: linux-btrfs@vger.kernel.org
Subject: newer /bin/cp have worse btrfs fiemap performance.
Date: Sun, 06 Nov 2022 07:30:29 +0800	[thread overview]
Message-ID: <20221106073028.71F9.409509F4@e16-tech.com> (raw)

Hi,

newer /bin/cp have worse btrfs fiemap performance.

btrfs version: misc-next 750de989d367(the lastest of 2022/11/05)
/bin/cp versions(run them on the same server)
    8.22 copy from centos/7.9
    8.30 rocklinux/9.0
    9.1 local build, https://kojipkgs.fedoraproject.org/packages/coreutils/9.1/8.eln121/

test case:
    /bin/cp /mnt/test/file1 /dev/null
    /mnt/test/file1 is created by https://lore.kernel.org/linux-btrfs/YuwUw2JLKtIa9X+S@localhost.localdomain/T/#T
    and /mnt/test/file1 is 256M.

    file is not cached: 'echo 3 >/proc/sys/vm/drop_caches'
    file is cached: run '/bin/cp /mnt/test/file1 /dev/null' again.

performance result(/bin/cp /mnt/test/file1 /dev/null):
/bin/cp 9.1
    file is not cached: 94.85(1:34.85)
    file is cached: 1982.43(33:02.43)
/bin/cp 8.30
    file is not cached: 1.48(0:01.48)
    file is cached:14.07(0:14.07)
/bin/cp 8.22
    file is not cached: 0.53(0:00.53)
    file is cached: 0.10(0:00.10)

as a compare, we test it on xfs too.
1) /bin/cp 8.22/8.30/9.1 have almost same performance.
2) the case(the file is cached) is faster than the case(the file is not
cached).

strace show that the logical of /bin/cp 8.30 and 9.1 are different.
/bin/cp 8.30
    lseek(3, 198737920, SEEK_SET)           = 198737920
    write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    read(3, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    write(4, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    lseek(3, 198746112, SEEK_SET)           = 198746112
    write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    read(3, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    write(4, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096

/bin/cp 9.1
    lseek(3, 880640, SEEK_DATA)             = 884736
    lseek(3, 884736, SEEK_HOLE)             = 888832
    lseek(3, 884736, SEEK_SET)              = 884736
    write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    read(3, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    write(4, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    lseek(3, 888832, SEEK_DATA)             = 892928
    lseek(3, 892928, SEEK_HOLE)             = 897024
    lseek(3, 892928, SEEK_SET)              = 892928
    write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    read(3, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096
    write(4, "a\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4096

Do we need some job in btrfs to support /bin/cp 9.1 well,
or /bin/cp 9.1 is just wrong?

Best Regards
Wang Yugui (wangyugui@e16-tech.com)
2022/11/06



             reply	other threads:[~2022-11-05 23:30 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-05 23:30 Wang Yugui [this message]
2022-11-06  0:16 ` newer /bin/cp have worse btrfs fiemap performance Wang Yugui
2022-11-07  8:36 ` Filipe Manana

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=20221106073028.71F9.409509F4@e16-tech.com \
    --to=wangyugui@e16-tech.com \
    --cc=linux-btrfs@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