Re: [PATCH 1/2] xfs: test the xfs_db path command

["Darrick J. Wong" <djwong@kernel.org>]

On Thu, Mar 25, 2021 at 03:03:14PM +0530, Chandan Babu R wrote:
> On 23 Mar 2021 at 09:50, Darrick J. Wong wrote:
> > From: Darrick J. Wong <djwong@kernel.org>
> >
> > Add a new test to make sure the xfs_db path command works the way the
> > author thinks it should.
> >
> > Signed-off-by: Darrick J. Wong <djwong@kernel.org>
> > ---
> >  tests/xfs/917     |   98 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/xfs/917.out |   19 ++++++++++
> >  tests/xfs/group   |    1 +
> >  3 files changed, 118 insertions(+)
> >  create mode 100755 tests/xfs/917
> >  create mode 100644 tests/xfs/917.out
> >
> >
> > diff --git a/tests/xfs/917 b/tests/xfs/917
> > new file mode 100755
> > index 00000000..bf21b290
> > --- /dev/null
> > +++ b/tests/xfs/917
> > @@ -0,0 +1,98 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0-or-later
> > +# Copyright (c) 2021 Oracle.  All Rights Reserved.
> > +#
> > +# FS QA Test No. 917
> > +#
> > +# Make sure the xfs_db path command works the way the author thinks it does.
> > +# This means that it can navigate to random inodes, fails on paths that don't
> > +# resolve.
> > +#
> > +seq=`basename $0`
> > +seqres=$RESULT_DIR/$seq
> > +echo "QA output created by $seq"
> > +
> > +here=`pwd`
> > +tmp=/tmp/$$
> > +status=1    # failure is the default!
> > +trap "_cleanup; exit \$status" 0 1 2 3 15
> > +
> > +_cleanup()
> > +{
> > +	cd /
> > +	rm -f $tmp.*
> > +}
> > +
> > +# get standard environment, filters and checks
> > +. ./common/rc
> > +. ./common/filter
> > +
> > +# real QA test starts here
> > +_supported_fs xfs
> > +_require_xfs_db_command "path"
> > +_require_scratch
> > +
> > +echo "Format filesystem and populate"
> > +_scratch_mkfs > $seqres.full
> > +_scratch_mount >> $seqres.full
> > +
> > +mkdir $SCRATCH_MNT/a
> > +mkdir $SCRATCH_MNT/a/b
> > +$XFS_IO_PROG -f -c 'pwrite 0 61' $SCRATCH_MNT/a/c >> $seqres.full
> > +ln -s -f c $SCRATCH_MNT/a/d
> > +mknod $SCRATCH_MNT/a/e b 8 0
> > +ln -s -f b $SCRATCH_MNT/a/f
> 
> Later in the test script, there are two checks corresponding to accessibility
> of file symlink and dir symlink. However, $SCRATCH_MNT/a/d and
> $SCRATCH_MNT/a/f are actually referring to non-existant files since current
> working directory at the time of invocation of ln command is the xfstests
> directory.
> 
> i.e. 'c' and 'b' arguments to 'ln' command above must be qualified with
> $SCRATCH_MNT/a/.

Hm?  d and f look fine to me:

$ ./check xfs/917
$ mount /dev/sdf /opt
$ cd /opt/a
$ ls
total 4
drwxr-xr-x 2 root root    6 Mar 25 09:25 b/
-rw------- 1 root root   61 Mar 25 09:25 c
lrwxrwxrwx 1 root root    1 Mar 25 09:25 d -> c
brw-r--r-- 1 root root 8, 0 Mar 25 09:25 e
lrwxrwxrwx 1 root root    1 Mar 25 09:25 f -> b/

The link target is copied verbatim into the symlink, so I don't see why
they need to be qualified?

(FWIW the path command doesn't resolve symlinks, so it really only
checks that /a/d and /a/f exist and are of type symlink.)

--D

> 
> > +
> > +_scratch_unmount
> > +
> > +echo "Check xfs_db path on directories"
> > +_scratch_xfs_db -c 'path /a' -c print | grep -q 'sfdir.*count.* 5$' || \
> > +	echo "Did not find directory /a"
> > +
> > +_scratch_xfs_db -c 'path /a/b' -c print | grep -q sfdir || \
> > +	echo "Did not find empty sf directory /a/b"
> > +
> > +echo "Check xfs_db path on files"
> > +_scratch_xfs_db -c 'path /a/c' -c print | grep -q 'core.size.*61' || \
> > +	echo "Did not find 61-byte file /a/c"
> > +
> > +echo "Check xfs_db path on file symlinks"
> > +_scratch_xfs_db -c 'path /a/d' -c print | grep -q symlink || \
> > +	echo "Did not find symlink /a/d"
> > +
> > +echo "Check xfs_db path on bdevs"
> > +_scratch_xfs_db -c 'path /a/e' -c print | grep -q 'format.*dev' || \
> > +	echo "Did not find bdev /a/e"
> > +
> > +echo "Check xfs_db path on dir symlinks"
> > +_scratch_xfs_db -c 'path /a/f' -c print | grep -q symlink || \
> > +	echo "Did not find symlink /a/f"
> 
> --
> chandan