From: "Darrick J. Wong" <djwong@kernel.org>
To: "liuyd.fnst@fujitsu.com" <liuyd.fnst@fujitsu.com>
Cc: "guaneryu@gmail.com" <guaneryu@gmail.com>,
"zlang@redhat.com" <zlang@redhat.com>,
"linux-xfs@vger.kernel.org" <linux-xfs@vger.kernel.org>,
"fstests@vger.kernel.org" <fstests@vger.kernel.org>,
"guan@eryu.me" <guan@eryu.me>
Subject: Re: [PATCH 1/9] seek_sanity_test: fix allocation unit detection on XFS realtime
Date: Wed, 27 Jul 2022 19:31:24 -0700 [thread overview]
Message-ID: <YuH0/H23l6jGsuNd@magnolia> (raw)
In-Reply-To: <3f63e720-c252-a836-b700-7a5739312b1b@fujitsu.com>
On Thu, Jul 28, 2022 at 01:37:08AM +0000, liuyd.fnst@fujitsu.com wrote:
> Hi, guys.
>
> Recently I hit a regression during test xfstest. Reverting this commit
> could fix that error.
>
> Reproduce steps (only nfs4.2 has this error):
> ```
> # cat local.config
>
> export TEST_DEV=127.0.0.1:/home/nfs/share0
>
> export TEST_DIR=/mnt/test
>
> export SCRATCH_DEV=127.0.0.1:/home/nfs/share1
ahahaaaa, NFS. I forgot that it sets st_blksize == 1048576, which will
trip this up.
Let me work on a fix to seek_sanity_test.c to make this work for xfs
realtime without breaking NFS. I /think/ the solution is to replace the
"if (...) goto done;" bit with something that queries FS_IOC_GETXATTR
and XFS_IOC_FSGEOMETRY to figure out if the file being tested is a
realtime file on an XFS filesystem, and set alloc_size to the rt extent
size.
/* Compute the file allocation unit size for an XFS file. */
static int detect_xfs_alloc_unit(int fd)
{
struct fsxattr fsx;
struct xfs_fsop_geom fsgeom;
int ret;
ret = ioctl(fd, XFS_IOC_FSGEOMETRY, &fsgeom);
if (ret)
return -1;
ret = ioctl(fd, FS_IOC_FSGETXATTR, &fsx);
if (ret)
return -1;
alloc_size = fsgeom.blocksize;
if (fsx.fsx_xflags & FS_XFLAG_REALTIME)
alloc_size *= fsgeom.rtextsize;
return 0;
}
should suffice to fix these testcase on xfs realtime without messing up
nfs.
--D
> export SCRATCH_MNT=/mnt/scratch
>
> export FSX_AVOID="-E"
>
> export NFS_MOUNT_OPTIONS="-o rw,relatime,vers=4.2"
>
>
> # ./check -nfs generic/285
>
> FSTYP -- nfs
>
> PLATFORM -- Linux/aarch64 hpe-apollo80-01-n00
> 5.14.0-131.el9.aarch64 #1 SMP PREEMPT_DYNAMIC Mon Jul 18 16:13:44 EDT 2022
>
> MKFS_OPTIONS -- 127.0.0.1:/home/nfs/share1
>
> MOUNT_OPTIONS -- -o rw,relatime,vers=4.2 -o
> context=system_u:object_r:root_t:s0 127.0.0.1:/home/nfs/share1 /mnt/scratch
>
>
>
> generic/285 2s ... [failed, exit status 1]- output mismatch (see
> /root/xfstests/results//generic/285.out.bad)
>
> --- tests/generic/285.out 2022-07-27 21:07:43.160268552 -0400
>
> +++ /root/xfstests/results//generic/285.out.bad 2022-07-27
> 21:31:27.887090532 -0400
>
> @@ -1 +1,3 @@
>
> QA output created by 285
>
> +seek sanity check failed!
>
> +(see /root/xfstests/results//generic/285.full for details)
>
> ...
>
> (Run 'diff -u /root/xfstests/tests/generic/285.out
> /root/xfstests/results//generic/285.out.bad' to see the entire diff)
>
> Ran: generic/285
>
> Failures: generic/285
>
> Failed 1 of 1 tests
>
>
> ```
>
> Reverting this commit then test pass.
> ```
> # git revert e861a30255c9780425ee5193325d30882fbe7410
> # make -j && make install -j
> ---snip---
> # ./check -nfs generic/285
>
> FSTYP -- nfs
>
> PLATFORM -- Linux/aarch64 hpe-apollo80-01-n00
> 5.14.0-131.el9.aarch64 #1 SMP PREEMPT_DYNAMIC Mon Jul 18 16:13:44 EDT 2022
>
> MKFS_OPTIONS -- 127.0.0.1:/home/nfs/share1
>
> MOUNT_OPTIONS -- -o rw,relatime,vers=4.2 -o
> context=system_u:object_r:root_t:s0 127.0.0.1:/home/nfs/share1 /mnt/scratch
>
>
>
> generic/285 1s ... 1s
>
> Ran: generic/285
>
> Passed all 1 tests
>
> ```
>
> On 6/29/22 04:21, Darrick J. Wong wrote:
> > From: Darrick J. Wong <djwong@kernel.org>
> >
> > The seek sanity test tries to figure out a file space allocation unit by
> > calling stat and then using an iterative SEEK_DATA method to try to
> > detect a smaller blocksize based on SEEK_DATA's consultation of the
> > filesystem's internal block mapping. This was put in (AFAICT) because
> > XFS' stat implementation returns max(filesystem blocksize, PAGESIZE) for
> > most regular files.
> >
> > Unfortunately, for a realtime file with an extent size larger than a
> > single filesystem block this doesn't work at all because block mappings
> > still work at filesystem block granularity, but allocation units do not.
> > To fix this, detect the specific case where st_blksize != PAGE_SIZE and
> > trust the fstat results.
> >
> > Signed-off-by: Darrick J. Wong <djwong@kernel.org>
> > ---
> > src/seek_sanity_test.c | 12 +++++++++++-
> > 1 file changed, 11 insertions(+), 1 deletion(-)
> >
> >
> > diff --git a/src/seek_sanity_test.c b/src/seek_sanity_test.c
> > index 76587b7f..1030d0c5 100644
> > --- a/src/seek_sanity_test.c
> > +++ b/src/seek_sanity_test.c
> > @@ -45,6 +45,7 @@ static int get_io_sizes(int fd)
> > off_t pos = 0, offset = 1;
> > struct stat buf;
> > int shift, ret;
> > + int pagesz = sysconf(_SC_PAGE_SIZE);
> >
> > ret = fstat(fd, &buf);
> > if (ret) {
> > @@ -53,8 +54,16 @@ static int get_io_sizes(int fd)
> > return ret;
> > }
> >
> > - /* st_blksize is typically also the allocation size */
> > + /*
> > + * st_blksize is typically also the allocation size. However, XFS
> > + * rounds this up to the page size, so if the stat blocksize is exactly
> > + * one page, use this iterative algorithm to see if SEEK_DATA will hint
> > + * at a more precise answer based on the filesystem's (pre)allocation
> > + * decisions.
> > + */
> > alloc_size = buf.st_blksize;
> > + if (alloc_size != pagesz)
> > + goto done;
> >
> > /* try to discover the actual alloc size */
> > while (pos == 0 && offset < alloc_size) {
> > @@ -80,6 +89,7 @@ static int get_io_sizes(int fd)
> > if (!shift)
> > offset += pos ? 0 : 1;
> > alloc_size = offset;
> > +done:
> > fprintf(stdout, "Allocation size: %ld\n", alloc_size);
> > return 0;
> >
> >
next prev parent reply other threads:[~2022-07-28 2:31 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-06-28 20:21 [PATCHSET 0/9] fstests: random fixes Darrick J. Wong
2022-06-28 20:21 ` [PATCH 1/9] seek_sanity_test: fix allocation unit detection on XFS realtime Darrick J. Wong
2022-06-29 7:41 ` Christoph Hellwig
2022-07-28 1:37 ` liuyd.fnst
2022-07-28 2:31 ` Darrick J. Wong [this message]
2022-07-28 2:37 ` Darrick J. Wong
2022-07-28 3:38 ` liuyd.fnst
2022-07-28 3:54 ` Darrick J. Wong
2022-06-28 20:21 ` [PATCH 2/9] xfs/070: filter new superblock verifier messages Darrick J. Wong
2022-06-29 4:15 ` Dave Chinner
2022-06-29 9:12 ` Zorro Lang
2022-06-29 22:41 ` Darrick J. Wong
2022-06-30 2:18 ` Zorro Lang
2022-06-28 20:21 ` [PATCH 3/9] xfs: test mkfs.xfs sizing of internal logs that Darrick J. Wong
2022-06-29 4:18 ` Dave Chinner
2022-06-29 22:43 ` Darrick J. Wong
2022-06-30 6:37 ` Zorro Lang
2022-06-30 19:32 ` Darrick J. Wong
2022-06-28 20:21 ` [PATCH 4/9] xfs: test xfs_copy doesn't do cached read before libxfs_mount Darrick J. Wong
2022-06-29 4:20 ` Dave Chinner
2022-06-29 22:46 ` Darrick J. Wong
2022-06-30 0:48 ` [PATCH v2.1 " Darrick J. Wong
2022-07-01 1:06 ` Dave Chinner
2022-06-28 20:21 ` [PATCH 5/9] check: document mkfs.xfs reliance on fstests exports Darrick J. Wong
2022-06-29 4:22 ` Dave Chinner
2022-06-28 20:21 ` [PATCH 6/9] xfs/109: handle larger minimum filesystem size Darrick J. Wong
2022-06-29 4:22 ` Dave Chinner
2022-06-28 20:21 ` [PATCH 7/9] xfs/018: fix LARP testing for small block sizes Darrick J. Wong
2022-06-28 20:22 ` [PATCH 8/9] xfs/166: fix golden output failures when multipage folios enabled Darrick J. Wong
2022-06-29 4:23 ` Dave Chinner
2022-06-28 20:22 ` [PATCH 9/9] xfs/547: fix problems with realtime Darrick J. Wong
2022-07-05 14:28 ` [PATCHSET 0/9] fstests: random fixes Zorro Lang
2022-07-05 22:04 ` Darrick J. Wong
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=YuH0/H23l6jGsuNd@magnolia \
--to=djwong@kernel.org \
--cc=fstests@vger.kernel.org \
--cc=guan@eryu.me \
--cc=guaneryu@gmail.com \
--cc=linux-xfs@vger.kernel.org \
--cc=liuyd.fnst@fujitsu.com \
--cc=zlang@redhat.com \
/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