Re: 8.0beta : missing umount at end of tests

[Jim Meyering]

Gilles Espinasse wrote:
> Selon Jim Meyering <address@hidden>:
>> Gilles Espinasse wrote:
>> ...
>> > The problem could be shorten to
>> >> mount --bind /dev/shm/tmp10767 a/b
>> > ...
>> >> + umount /dev/shm/tmp10767
>> >> umount: /dev/shm/tmp10767: not mounted
>> >
>> > mount think a/b is mounted, not /dev/shm/tmp10767
>> >
>> >
>> > If that matter
>> > [chroot-i486] root:/usr/src/coreutils-8.0$ mount --version
>> > mount from util-linux-ng 2.16.1 (with libblkid support)
>> >
>> > [chroot-i486] root:/usr/src/coreutils-8.0$ ls -l /etc/mtab
>> > lrwxrwxrwx 1 root root 12 Oct 18 09:42 /etc/mtab -> /proc/mounts
>>
>> Do you think this is an artifact of your using LFS
>> or from some aspect of your chroot?
>>
> I really don't know.
> IPCop script are mostly similar to LFS jhalf automated scripts but not exactly
> the same in some details. Our master script is a bash script and child scripts
> are makefile when it's exactly the opposite in jhalf.
> I will check first if this happen in LFS jhalf.
>
> And I will test the fix.
>
>> In any case, this should fix it.
>> Please confirm:
>>
>> diff --git a/tests/rm/one-file-system b/tests/rm/one-file-system
>> index c8e11eb..d1821c5 100755
>> --- a/tests/rm/one-file-system
>> +++ b/tests/rm/one-file-system
>> @@ -28,6 +28,8 @@ require_root_
>>  cleanup_()
>>  {
>>    umount "$other_partition_tmpdir"
>> +  # On some systems, the above fails, yet unmounting a/b works, so...
>> +  umount a/b
>>    rm -rf "$other_partition_tmpdir"
>>  }
>>  . "$abs_srcdir/other-fs-tmpdir"
>>
>> If that doesn't help, please show me what df -h prints,
>> or just send the relevant line from /proc/mounts.
>>
> Thank for your help

Please let me know, if/when you test it, preferably with a few
details about your system so I can justify this work-around.