Am 05.12.2010 21:49, schrieb Nix:
> On 5 Dec 2010, Jakob Unterwurzacher verbalised:
>
>> Am 05.12.2010 18:07, schrieb Nix:
>>> mount --move doesn't work as specified in the manpage:
>>>
>>> ,----
>>> | # cd /tmp
>>> | # mkdir foo
>>> | # mkdir bar
>>> | # mount --move foo bar
>>> | mount: wrong fs type, bad option, bad superblock on /tmp/foo,
>>> | missing codepage or helper program, or other error
>>> | In some cases useful info is found in syslog - try
>>> | dmesg | tail or so
>>
>> man page says:
>>
>> The move operation.
>> Since Linux 2.5.1 it is possible to atomically move a
>> mounted tree to another place. The call is
>> mount --move olddir newdir
>>
>> The magic word is *mounted* I'm afraid, you can only move mountpoints.
>
> Oh, that's rather crappy. This doesn't apply to MS_BIND, which is
> strange: are you sure this is actually a kernel restriction?
Yes,
# strace mount --move foo bar
gives:
mount("/tmp/foo", "bar", 0x805577c, MS_MGC_VAL|MS_MOVE, NULL) = -1
EINVAL (Invalid argument)
-> The kernel says no.
>> cd /tmp
>> mkdir foo
>> mount --bind foo foo
>> mkdir bar
>> mount --move foo bar
>
> I might just as well do a mount --bind in that case, which *can* be used
> on absolutely arbitrary directories (and indeed individual files).
>
>> Though it does not do what you want as you will see foo's content in foo
>> AND bar.
>
> That's what I'm trying to avoid. I thought MS_MOVE and MS_BIND together
> allowed this sort of cut-and-paste: apparently not. :/
>
> I can probably get away with it by mounting a size-zero tmpfs over the
> original mount point after the bind operation. But, still, ew.
If you are *not* dealing with mountpoints, isn't renaming the directory
enough?
--
To unsubscribe from this list: send the line "unsubscribe util-linux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
