Source: bup
Version: 0.25-1

It failed to build on arm64:

http://buildd.debian.org/status/package.php?p=bup&suite=sid

The warnings and error were:

! t/test.sh  save --strip-path (no match): ............
df: Warning: cannot read table of mounted file systems: No such file
or directory
df: Warning: cannot read table of mounted file systems: No such file
or directory
df: Warning: cannot read table of mounted file systems: No such file
or directory
df: Warning: cannot read table of mounted file systems: No such file
or directory
! t/test.sh:466  path-filesystems .                                 ok
! t/test.sh:466  sort -u                                            ok
! t/test.sh:466  wc -l                                              ok
! t/test.sh:477  force-delete graft-points.tmp                      ok
! t/test.sh:478  mkdir graft-points.tmp                             ok
! t/test.sh:479  pwd                                                ok
Initialized empty Git repository in /«PKGBUILDDIR»/graft-points.tmp/bup/
! t/test.sh:480  bup init                                           ok
! t/test.sh:481  mkdir -p graft-points.tmp/src/x/y/z                ok
! t/test.sh:482  bup random 8k                                      ok
! t/test.sh:483  bup random 8k                                      ok
! t/test.sh:484  bup index -u graft-points.tmp/src                  ok
! t/test.sh:485  bup save --strip-path graft-points.tmp/foo -n foo
graft-points.tmp/src/x ok
Linux chattr: [Errno 25] Inappropriate ioctl for device: '.' (0x0x80000)
WARNING: 1 errors encountered while restoring.
! t/test.sh:486  bup restore -C graft-points.tmp/restore /foo/latest FAILED

I can't reproduce this error. It works when I build it.

However, exactly the same failure has happened several times on a buildd.

The test that fails is this one in t/test.sh:

WVSTART "save --strip-path (no match)"
(
    if test $(WVPASS path-filesystems . | WVPASS sort -u | WVPASS wc -l) -ne 1
    then
        # Skip the test because the attempt to restore parent dirs to
        # the current filesystem may fail -- i.e. running from
        # /foo/ext4/bar/btrfs will fail when bup tries to restore
        # linux attrs above btrfs to the restore tree *inside* btrfs.
        echo "(running from tree with mixed filesystems; skipping test)" 1>&2
        exit 0
    fi

    tmp=graft-points.tmp
    WVPASS force-delete $tmp
    WVPASS mkdir $tmp
    export BUP_DIR="$(WVPASS pwd)/$tmp/bup" || exit $?
    WVPASS bup init
    WVPASS mkdir -p $tmp/src/x/y/z
    WVPASS bup random 8k > $tmp/src/x/y/random-1
    WVPASS bup random 8k > $tmp/src/x/y/z/random-2
    WVPASS bup index -u $tmp/src
    WVPASS bup save --strip-path $tmp/foo -n foo $tmp/src/x
    WVPASS bup restore -C $tmp/restore /foo/latest
    WVPASS t/compare-trees $tmp/src/ "$tmp/restore/latest/$TOP/$tmp/src/"
) || exit $?

The "Warning" comes from path-filesystems doing "df -T .". It probably
doesn't matter as it's only used for detecting "mixed filesystems",
which we presumably don't have on the buildd.

The error comes from the "bup restore". From looking at the ioctls on
a successful run it seems that bup changes its working directory to
"/" and then calls bup_set_linux_file_attr on a sequence of
directories finishing at ".":

build/deb/bup/bup-0.25/graft-points.tmp/src/x/y/z/random-2
build/deb/bup/bup-0.25/graft-points.tmp/src/x/y/z
...
build/deb
build
.

So it seems that the ioctl fails on the root directory on the arm64
buildd, for some reason, but not on the buildds for other
architectures.

It appears that this test, which is failing, has been removed in
bup-0.26, which may explain why 0.26~rc1-1 was successfully built,
though that was a different buildd, so that would be an alternative
explanation.

If it's not required (why was it removed in 0.26?), please consider
removing this test from the version of bup in sid.

Or perhaps the information I've given above will help someone to
understand why this test is failing on arm-arm-01. (I'd like to see
what happens if arm-linaro-* tries it.)


--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to