Linux-Block Archive on lore.kernel.org
 help / color / Atom feed
From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: Christoph Hellwig <hch@lst.de>
Cc: linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	linux-block@vger.kernel.org, linux-mm@kvack.org
Subject: Re: buffered I/O without buffer heads in xfs and iomap v2
Date: Mon, 21 May 2018 13:46:10 -0700
Message-ID: <20180521204610.GC4507@magnolia> (raw)
In-Reply-To: <20180518164830.1552-1-hch@lst.de>

On Fri, May 18, 2018 at 06:47:56PM +0200, Christoph Hellwig wrote:
> Hi all,
> 
> this series adds support for buffered I/O without buffer heads to
> the iomap and XFS code.
> 
> For now this series only contains support for block size == PAGE_SIZE,
> with the 4k support split into a separate series.
> 
> 
> A git tree is available at:
> 
>     git://git.infradead.org/users/hch/xfs.git xfs-iomap-read.2
> 
> Gitweb:
> 
>     http://git.infradead.org/users/hch/xfs.git/shortlog/refs/heads/xfs-iomap-read.2

Hmm, so I pulled this and ran my trivial stupid benchmark on for-next.
It's a stupid VM with a 2G of RAM and a 12GB virtio-scsi disk backed by
tmpfs:

# mkfs.xfs -f -m rmapbt=0,reflink=1 /dev/sda
meta-data=/dev/sda               isize=512    agcount=4, agsize=823296
blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1        finobt=1, sparse=1,
rmapbt=1
         =                       reflink=1
data     =                       bsize=4096   blocks=3293184, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0, ftype=1
log      =internal log           bsize=4096   blocks=3693, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0
# mount /dev/sda /mnt
# xfs_io -f -c 'pwrite -W -S 0x64 -b 83886080 0 734003200' /mnt/a
wrote 734003200/734003200 bytes at offset 0
700 MiB, 9 ops; 0:00:01.06 (655.500 MiB/sec and 8.4279 ops/sec)
# cp --reflink=always /mnt/a /mnt/b
# xfs_io -f -c 'pwrite -W -S 0x65 -b 83886080 0 734003200' /mnt/b
wrote 734003200/734003200 bytes at offset 0
700 MiB, 9 ops; 0.9620 sec (727.615 MiB/sec and 9.3551 ops/sec)

Then I applied your series (not including the blocksize < pagesize
series) and saw this big regression:

# mkfs.xfs -f -m rmapbt=0,reflink=1 /dev/sda
meta-data=/dev/sda               isize=512    agcount=4, agsize=823296
blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1        finobt=1, sparse=1,
rmapbt=1
         =                       reflink=1
data     =                       bsize=4096   blocks=3293184, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0, ftype=1
log      =internal log           bsize=4096   blocks=3693, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0
# mount /dev/sda /mnt
# xfs_io -f -c 'pwrite -W -S 0x64 -b 83886080 0 734003200' /mnt/a
wrote 734003200/734003200 bytes at offset 0
700 MiB, 9 ops; 0:00:08.04 (87.031 MiB/sec and 1.1190 ops/sec)
# cp --reflink=always /mnt/a /mnt/b
# xfs_io -f -c 'pwrite -W -S 0x65 -b 83886080 0 734003200' /mnt/b
wrote 734003200/734003200 bytes at offset 0
700 MiB, 9 ops; 0:00:21.61 (32.389 MiB/sec and 0.4164 ops/sec)

I'll see if I can spot the problem while I read through the v2 code...

--D

> 
> Changes since v1:
>  - fix the iomap_readpages error handling
>  - use unsigned file offsets in a few places to avoid arithmetic overflows
>  - allocate a iomap_page in iomap_page_mkwrite to fix generic/095
>  - improve a few comments
>  - add more asserts
>  - warn about truncated block numbers from ->bmap
>  - new patch to change the __do_page_cache_readahead return value to
>    unsigned int
>  - remove an incorrectly added empty line
>  - make inline data an explicit iomap type instead of a flag
>  - add a IOMAP_F_BUFFER_HEAD flag to force use of buffers heads for gfs2,
>    and keep the basic buffer head infrastructure around for now.

  parent reply index

Thread overview: 49+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-05-18 16:47 Christoph Hellwig
2018-05-18 16:47 ` [PATCH 01/34] block: add a lower-level bio_add_page interface Christoph Hellwig
2018-05-18 16:56   ` Jens Axboe
2018-05-18 16:47 ` [PATCH 02/34] fs: factor out a __generic_write_end helper Christoph Hellwig
2018-05-18 16:47 ` [PATCH 03/34] fs: move page_cache_seek_hole_data to iomap.c Christoph Hellwig
2018-05-18 16:48 ` [PATCH 04/34] fs: remove the buffer_unwritten check in page_seek_hole_data Christoph Hellwig
2018-05-18 16:48 ` [PATCH 05/34] fs: use ->is_partially_uptodate in page_cache_seek_hole_data Christoph Hellwig
2018-05-21 19:53   ` Darrick J. Wong
2018-05-22  7:56     ` Christoph Hellwig
2018-05-18 16:48 ` [PATCH 06/34] mm: give the 'ret' variable a better name __do_page_cache_readahead Christoph Hellwig
2018-05-18 16:48 ` [PATCH 07/34] mm: return an unsigned int from __do_page_cache_readahead Christoph Hellwig
2018-05-18 16:48 ` [PATCH 08/34] mm: split ->readpages calls to avoid non-contiguous pages lists Christoph Hellwig
2018-05-22 14:52   ` Brian Foster
2018-05-18 16:48 ` [PATCH 09/34] iomap: inline data should be an iomap type, not a flag Christoph Hellwig
2018-05-18 16:48 ` [PATCH 10/34] iomap: fix the comment describing IOMAP_NOWAIT Christoph Hellwig
2018-05-18 16:48 ` [PATCH 11/34] iomap: move IOMAP_F_BOUNDARY to gfs2 Christoph Hellwig
2018-05-18 16:48 ` [PATCH 12/34] iomap: use __bio_add_page in iomap_dio_zero Christoph Hellwig
2018-05-18 16:48 ` [PATCH 13/34] iomap: add a iomap_sector helper Christoph Hellwig
2018-05-18 16:48 ` [PATCH 14/34] iomap: add an iomap-based bmap implementation Christoph Hellwig
2018-05-18 16:48 ` [PATCH 15/34] iomap: add an iomap-based readpage and readpages implementation Christoph Hellwig
2018-05-18 16:48 ` [PATCH 16/34] iomap: add initial support for writes without buffer heads Christoph Hellwig
2018-05-21 23:27   ` Darrick J. Wong
2018-05-22  0:07     ` Dave Chinner
2018-05-22  8:24       ` Christoph Hellwig
2018-05-22 22:38         ` Dave Chinner
2018-05-23  6:03           ` Christoph Hellwig
2018-05-22  8:20     ` Christoph Hellwig
2018-05-22  8:31     ` Christoph Hellwig
2018-05-22  9:39       ` Christoph Hellwig
2018-05-18 16:48 ` [PATCH 17/34] xfs: use iomap_bmap Christoph Hellwig
2018-05-18 16:48 ` [PATCH 18/34] xfs: use iomap for blocksize == PAGE_SIZE readpage and readpages Christoph Hellwig
2018-05-18 16:48 ` [PATCH 19/34] xfs: simplify xfs_bmap_punch_delalloc_range Christoph Hellwig
2018-05-18 16:48 ` [PATCH 20/34] xfs: simplify xfs_aops_discard_page Christoph Hellwig
2018-05-18 16:48 ` [PATCH 21/34] xfs: move locking into xfs_bmap_punch_delalloc_range Christoph Hellwig
2018-05-18 16:48 ` [PATCH 22/34] xfs: make xfs_writepage_map extent map centric Christoph Hellwig
2018-05-18 16:48 ` [PATCH 23/34] xfs: remove the now unused XFS_BMAPI_IGSTATE flag Christoph Hellwig
2018-05-18 16:48 ` [PATCH 24/34] xfs: remove xfs_reflink_find_cow_mapping Christoph Hellwig
2018-05-18 16:48 ` [PATCH 25/34] xfs: remove xfs_reflink_trim_irec_to_next_cow Christoph Hellwig
2018-05-18 16:48 ` [PATCH 26/34] xfs: simplify xfs_map_blocks by using xfs_iext_lookup_extent directly Christoph Hellwig
2018-05-18 16:48 ` [PATCH 27/34] xfs: don't clear imap_valid for a non-uptodate buffers Christoph Hellwig
2018-05-18 16:48 ` [PATCH 28/34] xfs: remove the imap_valid flag Christoph Hellwig
2018-05-18 16:48 ` [PATCH 29/34] xfs: don't look at buffer heads in xfs_add_to_ioend Christoph Hellwig
2018-05-18 16:48 ` [PATCH 30/34] xfs: move all writeback buffer_head manipulation into xfs_map_at_offset Christoph Hellwig
2018-05-18 16:48 ` [PATCH 31/34] xfs: remove xfs_start_page_writeback Christoph Hellwig
2018-05-18 16:48 ` [PATCH 32/34] xfs: refactor the tail of xfs_writepage_map Christoph Hellwig
2018-05-18 16:48 ` [PATCH 33/34] xfs: do not set the page uptodate in xfs_writepage_map Christoph Hellwig
2018-05-18 16:48 ` [PATCH 34/34] xfs: allow writeback on pages without buffer heads Christoph Hellwig
2018-05-21 20:46 ` Darrick J. Wong [this message]
2018-05-22 20:33   ` buffered I/O without buffer heads in xfs and iomap v2 Christoph Hellwig

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=20180521204610.GC4507@magnolia \
    --to=darrick.wong@oracle.com \
    --cc=hch@lst.de \
    --cc=linux-block@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-xfs@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

Linux-Block Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-block/0 linux-block/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-block linux-block/ https://lore.kernel.org/linux-block \
		linux-block@vger.kernel.org
	public-inbox-index linux-block

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-block


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git