From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: "Theodore Ts'o" <tytso@mit.edu>
Cc: linux-ext4@vger.kernel.org
Subject: Re: [PATCH 15/34] libext2fs: support BLKZEROOUT/FALLOC_FL_ZERO_RANGE in ext2fs_zero_blocks
Date: Mon, 20 Oct 2014 16:37:56 -0700 [thread overview]
Message-ID: <20141020233756.GA32084@birch.djwong.org> (raw)
In-Reply-To: <20141018163255.GB30124@thunk.org>
On Sat, Oct 18, 2014 at 12:32:55PM -0400, Theodore Ts'o wrote:
> On Sat, Sep 13, 2014 at 03:12:53PM -0700, Darrick J. Wong wrote:
> > Plumb a new call into the IO manager to support translating
> > ext2fs_zero_blocks calls into the equivalent kernel-level BLKZEROOUT
> > ioctl or FALLOC_FL_ZERO_RANGE fallocate flag primitives when possible.
> >
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> > contrib/fallocate.c | 14 +++++++++
>
> I've separated out the contrib/fallocate change and created a separate
> commit for it, since it really is a separate change.
>
> What I'd like to see for the zero_blocks change io_manager is:
>
> (a) if we try to zero a range past the end of the file, we should just
> truncate the file to set i_size. Similarly, if this is a regular
> file, we should try to use PUNCH_HOLE. We already try to keep a raw
> file system image file to be sparse, so I don't see any real problems
> with this.
Done. For files, it'll truncate if the file needs to be extended, and then
punch out the zero range. If punch isn't supported, it'll try zero-range
as a last resort.
> (b) for a block device, if IO_FLAG_DIRECT_IO is set, it shoud be safe
> to try to use te BLKZEROOUT. If not, we can use
> posix_fadvise(POSIX_FADV_DONTNEED) and verify that this correctly zaps
> the relevant parts of the buffer cache. If it doesn't do the right
> thing, we can use BLKFLSBUF, which will zap the entire buffer cache
> for the device. Which is pretty heavy weight, but I really think it
> only makes sense to use zeroout for zeroing the inode table and the
> journal file.
I agree that it makes sense not to zero-out single blocks on bdevs.
> Even if we patch the kernel to make BLKZEROOUT to automatically do
> this, we can't count on it, and in particular if it turns out we have
> to use BLKFLSBUF, we're not going to want to use this for zero'ing a
> single 4k block. It doesn't happen that often, and I don't think
> there will be much if any measurable difference in performance if we
> use WRITE SAME vs. WRITE for a small region.
>
> Does this make sense?
>
> - Ted
>
> P.S. Once we do this, when using mke2fs on a file, we should really
> use punch_hole and disable lazy_itable_init, to save I/O bandwidth on
> VM's running on cloud systems.
Ok. I think current mke2fs does this if device discard is turned on.
Curiously, it'll still zero the itable even if itable_zeroed == 1.
--D
next prev parent reply other threads:[~2014-10-20 23:38 UTC|newest]
Thread overview: 67+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-09-13 22:11 [PATCH 00/34] e2fsprogs Summer 2014 patchbomb, part 6 Darrick J. Wong
2014-09-13 22:11 ` [PATCH 01/34] e2fsck: offer to clear overlapping extents Darrick J. Wong
2014-09-19 1:45 ` Theodore Ts'o
2014-09-13 22:11 ` [PATCH 02/34] e2fsck: fix sliding the directory block down on bigalloc Darrick J. Wong
2014-09-19 1:45 ` Theodore Ts'o
2014-09-13 22:11 ` [PATCH 03/34] misc: zero s_jnl_blocks when adding journal online or removing external journal Darrick J. Wong
2014-09-19 1:45 ` Theodore Ts'o
2014-09-13 22:11 ` [PATCH 04/34] libext2fs: ext2fs_new_block2() should call alloc_block hook Darrick J. Wong
2014-09-13 22:11 ` [PATCH 05/34] debugfs: manage needs_recover feature when messing with the journal Darrick J. Wong
2014-09-19 6:01 ` Theodore Ts'o
2014-09-13 22:11 ` [PATCH 06/34] debugfs: add LIBINTL to debugfs link command Darrick J. Wong
2014-09-19 4:46 ` Theodore Ts'o
2014-10-17 21:07 ` Darrick J. Wong
2014-10-18 16:10 ` Theodore Ts'o
2014-09-13 22:11 ` [PATCH 07/34] ext2fs: add readahead method to improve scanning Darrick J. Wong
2014-09-19 16:15 ` Theodore Ts'o
2014-09-13 22:12 ` [PATCH 08/34] libext2fs/e2fsck: provide routines to read-ahead metadata Darrick J. Wong
2014-09-13 22:12 ` [PATCH 09/34] e2fsck: read-ahead metadata during passes 1, 2, and 4 Darrick J. Wong
2014-09-13 22:12 ` [PATCH 10/34] dumpe2fs: provide a machine-readable group-only mode Darrick J. Wong
2014-09-19 16:17 ` Theodore Ts'o
2014-09-13 22:12 ` [PATCH 11/34] dumpe2fs: output cleanup Darrick J. Wong
2014-09-19 16:22 ` Theodore Ts'o
2014-09-19 20:00 ` Darrick J. Wong
2014-10-13 18:04 ` Darrick J. Wong
2014-09-13 22:12 ` [PATCH 12/34] misc: move check_plausibility into a separate file Darrick J. Wong
2014-09-19 22:16 ` Theodore Ts'o
2014-09-13 22:12 ` [PATCH 13/34] misc: add plausibility checks to debugfs/tune2fs/dumpe2fs/e2fsck Darrick J. Wong
2014-09-19 23:00 ` Theodore Ts'o
2014-09-13 22:12 ` [PATCH 14/34] misc: use libmagic when libblkid can't identify something Darrick J. Wong
2014-09-21 5:29 ` Theodore Ts'o
2014-09-13 22:12 ` [PATCH 15/34] libext2fs: support BLKZEROOUT/FALLOC_FL_ZERO_RANGE in ext2fs_zero_blocks Darrick J. Wong
2014-09-22 2:51 ` Theodore Ts'o
2014-09-29 18:58 ` Darrick J. Wong
2014-10-14 2:58 ` Darrick J. Wong
2014-10-18 16:32 ` Theodore Ts'o
2014-10-20 23:37 ` Darrick J. Wong [this message]
2014-09-13 22:12 ` [PATCH 16/34] libext2fs/e2fsck: refactor everyone who writes zero blocks to disk Darrick J. Wong
2014-10-13 10:09 ` Theodore Ts'o
2014-10-13 17:09 ` Darrick J. Wong
2014-09-13 22:13 ` [PATCH 17/34] libext2fs: support allocating uninit blocks in bmap2() Darrick J. Wong
2014-10-13 14:35 ` Theodore Ts'o
2014-10-13 16:56 ` Darrick J. Wong
2014-10-13 18:34 ` Darrick J. Wong
2014-09-13 22:13 ` [PATCH 18/34] libext2fs: file IO routines should handle uninit blocks Darrick J. Wong
2014-09-13 22:13 ` [PATCH 19/34] resize2fs: convert fs to and from 64bit mode Darrick J. Wong
2014-09-14 17:34 ` TR Reardon
2014-09-14 17:50 ` Darrick J. Wong
2014-09-13 22:13 ` [PATCH 20/34] resize2fs: adjust reserved_gdt_blocks when changing group descriptor size Darrick J. Wong
2014-09-13 22:13 ` [PATCH 21/34] tests: test resize2fs 32->64 and 64->32bit conversion code Darrick J. Wong
2014-09-13 22:13 ` [PATCH 22/34] libext2fs: find inode goal when allocating blocks Darrick J. Wong
2014-09-13 22:13 ` [PATCH 23/34] libext2fs: find/alloc a range of empty blocks Darrick J. Wong
2014-09-13 22:13 ` [PATCH 24/34] libext2fs: add new hooks to support large allocations Darrick J. Wong
2014-09-13 22:14 ` [PATCH 25/34] libext2fs: implement fallocate Darrick J. Wong
2014-09-13 22:14 ` [PATCH 26/34] libext2fs: use fallocate for creating journals and hugefiles Darrick J. Wong
2014-09-13 22:14 ` [PATCH 27/34] debugfs: implement fallocate Darrick J. Wong
2014-09-13 22:14 ` [PATCH 28/34] tests: test debugfs punch command Darrick J. Wong
2014-09-19 16:26 ` Theodore Ts'o
2014-09-19 20:01 ` Darrick J. Wong
2014-09-13 22:14 ` [PATCH 30/34] fuse2fs: translate ACL structures Darrick J. Wong
2014-09-13 22:14 ` [PATCH 31/34] fuse2fs: handle 64-bit dates correctly Darrick J. Wong
2014-09-13 22:14 ` [PATCH 32/34] fuse2fs: implement fallocate Darrick J. Wong
2014-09-13 22:15 ` [PATCH 34/34] tests: enable using fuse2fs with metadata checksum test Darrick J. Wong
2014-09-14 17:19 ` [PATCH 35/34] e2fsck: free bh when descriptor block checksum fails Darrick J. Wong
2014-09-14 19:11 ` Eric Sandeen
2014-09-19 1:46 ` Theodore Ts'o
2014-09-18 19:09 ` [PATCH 36/34] misc: fix Coverity complaints Darrick J. Wong
2014-09-19 1:47 ` Theodore Ts'o
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=20141020233756.GA32084@birch.djwong.org \
--to=darrick.wong@oracle.com \
--cc=linux-ext4@vger.kernel.org \
--cc=tytso@mit.edu \
/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).