From: "Darrick J. Wong" <djwong@kernel.org>
To: John Garry <john.g.garry@oracle.com>
Cc: chandan.babu@oracle.com, dchinner@redhat.com, hch@lst.de,
viro@zeniv.linux.org.uk, brauner@kernel.org, jack@suse.cz,
linux-xfs@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-fsdevel@vger.kernel.org, catherine.hoang@oracle.com,
martin.petersen@oracle.com
Subject: Re: [PATCH v4 02/14] xfs: always tail align maxlen allocations
Date: Fri, 23 Aug 2024 09:31:49 -0700 [thread overview]
Message-ID: <20240823163149.GC865349@frogsfrogsfrogs> (raw)
In-Reply-To: <20240813163638.3751939-3-john.g.garry@oracle.com>
On Tue, Aug 13, 2024 at 04:36:26PM +0000, John Garry wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> When we do a large allocation, the core free space allocation code
> assumes that args->maxlen is aligned to args->prod/args->mod. hence
> if we get a maximum sized extent allocated, it does not do tail
> alignment of the extent.
>
> However, this assumes that nothing modifies args->maxlen between the
> original allocation context setup and trimming the selected free
> space extent to size. This assumption has recently been found to be
> invalid - xfs_alloc_space_available() modifies args->maxlen in low
> space situations - and there may be more situations we haven't yet
> found like this.
>
> Force aligned allocation introduces the requirement that extents are
> correctly tail aligned, resulting in this occasional latent
> alignment failure to be reclassified from an unimportant curiousity
> to a must-fix bug.
>
> Removing the assumption about args->maxlen allocations always being
> tail aligned is trivial, and should not impact anything because
> args->maxlen for inodes with extent size hints configured are
> already aligned. Hence all this change does it avoid weird corner
> cases that would have resulted in unaligned extent sizes by always
> trimming the extent down to an aligned size.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> Reviewed-by: "Darrick J. Wong" <djwong@kernel.org> [provisional on v1 series comment]
Still provisional -- neither the original patch author nor the submitter
have answered my question from June:
IOWs, we always trim rlen, unless there is no alignment (prod==1) or
rlen is less than mod. For a forcealign file, it should never be the
case that minlen < mod because we'll have returned ENOSPC, right?
--D
> Signed-off-by: John Garry <john.g.garry@oracle.com>
> ---
> fs/xfs/libxfs/xfs_alloc.c | 12 +++++-------
> 1 file changed, 5 insertions(+), 7 deletions(-)
>
> diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c
> index d559d992c6ef..bf08b9e9d9ac 100644
> --- a/fs/xfs/libxfs/xfs_alloc.c
> +++ b/fs/xfs/libxfs/xfs_alloc.c
> @@ -433,20 +433,18 @@ xfs_alloc_compute_diff(
> * Fix up the length, based on mod and prod.
> * len should be k * prod + mod for some k.
> * If len is too small it is returned unchanged.
> - * If len hits maxlen it is left alone.
> */
> -STATIC void
> +static void
> xfs_alloc_fix_len(
> - xfs_alloc_arg_t *args) /* allocation argument structure */
> + struct xfs_alloc_arg *args)
> {
> - xfs_extlen_t k;
> - xfs_extlen_t rlen;
> + xfs_extlen_t k;
> + xfs_extlen_t rlen = args->len;
>
> ASSERT(args->mod < args->prod);
> - rlen = args->len;
> ASSERT(rlen >= args->minlen);
> ASSERT(rlen <= args->maxlen);
> - if (args->prod <= 1 || rlen < args->mod || rlen == args->maxlen ||
> + if (args->prod <= 1 || rlen < args->mod ||
> (args->mod == 0 && rlen < args->prod))
> return;
> k = rlen % args->prod;
> --
> 2.31.1
>
>
next prev parent reply other threads:[~2024-08-23 16:31 UTC|newest]
Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-08-13 16:36 [PATCH v4 00/14] forcealign for xfs John Garry
2024-08-13 16:36 ` [PATCH v4 01/14] xfs: only allow minlen allocations when near ENOSPC John Garry
2024-08-23 16:28 ` Darrick J. Wong
2024-08-13 16:36 ` [PATCH v4 02/14] xfs: always tail align maxlen allocations John Garry
2024-08-23 16:31 ` Darrick J. Wong [this message]
2024-08-29 17:58 ` John Garry
2024-08-29 21:34 ` Darrick J. Wong
2024-08-13 16:36 ` [PATCH v4 03/14] xfs: simplify extent allocation alignment John Garry
2024-08-13 16:36 ` [PATCH v4 04/14] xfs: make EOF allocation simpler John Garry
2024-09-04 18:25 ` Ritesh Harjani
2024-09-05 7:51 ` John Garry
2024-08-13 16:36 ` [PATCH v4 05/14] xfs: introduce forced allocation alignment John Garry
2024-08-13 16:36 ` [PATCH v4 06/14] xfs: align args->minlen for " John Garry
2024-08-13 16:36 ` [PATCH v4 07/14] xfs: Introduce FORCEALIGN inode flag John Garry
2024-08-13 16:36 ` [PATCH v4 08/14] xfs: Update xfs_inode_alloc_unitsize() for forcealign John Garry
2024-08-13 16:36 ` [PATCH v4 09/14] xfs: Update xfs_setattr_size() " John Garry
2024-08-13 16:36 ` [PATCH v4 10/14] xfs: Do not free EOF blocks " John Garry
2024-08-13 16:36 ` [PATCH v4 11/14] xfs: Only free full extents " John Garry
2024-08-13 16:36 ` [PATCH v4 12/14] xfs: Unmap blocks according to forcealign John Garry
2024-08-23 16:35 ` Darrick J. Wong
2024-08-13 16:36 ` [PATCH v4 13/14] xfs: Don't revert allocated offset for forcealign John Garry
2024-08-13 16:36 ` [PATCH v4 14/14] xfs: Enable file data forcealign feature John Garry
2024-09-04 18:14 ` [PATCH v4 00/14] forcealign for xfs Ritesh Harjani
2024-09-04 23:20 ` Dave Chinner
2024-09-05 3:56 ` Ritesh Harjani
2024-09-05 6:33 ` Dave Chinner
2024-09-10 2:51 ` Ritesh Harjani
2024-09-16 6:33 ` Dave Chinner
2024-09-10 12:33 ` Ritesh Harjani
2024-09-16 7:03 ` Dave Chinner
2024-09-16 10:24 ` John Garry
2024-09-17 20:54 ` Darrick J. Wong
2024-09-17 23:34 ` Dave Chinner
2024-09-17 22:12 ` Dave Chinner
2024-09-18 7:59 ` John Garry
2024-09-23 2:57 ` Dave Chinner
2024-09-23 3:33 ` Christoph Hellwig
2024-09-23 8:16 ` John Garry
2024-09-23 12:07 ` Christoph Hellwig
2024-09-23 12:33 ` John Garry
2024-09-24 6:17 ` Christoph Hellwig
2024-09-24 9:48 ` John Garry
2024-11-29 11:36 ` John Garry
2024-09-23 8:00 ` John Garry
2024-09-05 10:15 ` John Garry
2024-09-05 21:47 ` Dave Chinner
2024-09-06 14:31 ` John Garry
2024-09-08 22:49 ` Dave Chinner
2024-09-09 16:18 ` John Garry
2024-09-16 5:25 ` Dave Chinner
2024-09-16 9:44 ` John Garry
2024-09-17 22:27 ` Dave Chinner
2024-09-18 10:12 ` John Garry
2024-11-14 12:48 ` Long Li
2024-11-14 16:22 ` John Garry
2024-11-14 20:07 ` Dave Chinner
2024-11-15 8:14 ` John Garry
2024-11-15 11:20 ` Long Li
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=20240823163149.GC865349@frogsfrogsfrogs \
--to=djwong@kernel.org \
--cc=brauner@kernel.org \
--cc=catherine.hoang@oracle.com \
--cc=chandan.babu@oracle.com \
--cc=dchinner@redhat.com \
--cc=hch@lst.de \
--cc=jack@suse.cz \
--cc=john.g.garry@oracle.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=martin.petersen@oracle.com \
--cc=viro@zeniv.linux.org.uk \
/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).