--- Begin Message ---
Subject: |
mv directory cross-filesystem where destination exists fails to remove dest with EISDIR |
Date: |
Mon, 01 Jul 2013 22:21:36 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20130514 Thunderbird/17.0.6 |
Hi,
I have found a bug which affects RHEL 5.9, RHEL 6.4, Fedora 18, and the
latest version of coreutils from the git repo this morning (1st July
2013) in exactly the same way, and yet does not affect Solaris 10's mv
command.
Test case:
f18 # mkdir /test /home/test
f18 # cp /etc/passwd /home/test
f18 # mv /home/test /
mv: overwrite ‘/test’? y
mv: inter-device move failed: ‘/home/test’ to ‘/test’; unable to remove
target: Is a directory
Actual results:
mv: inter-device move failed: ‘/home/test’ to ‘/test’; unable to remove
target: Is a directory
Expected results:
/home/test/passwd is copied to /test/passwd
I have determined that the problem can be "fixed" (made to behave the
same as Solaris) by editting src/copy.c as follows:
2176c2176
< if (unlink (dst_name) != 0 && errno != ENOENT)
---
if (unlink (dst_name) != 0 && errno != ENOENT && errno != EISDIR)
2267,2269c2267,2278
< error (0, errno, _("cannot create directory %s"),
< quote (dst_name));
< goto un_backup;
---
> if (errno == EEXIST)
> {
> if (lchmod (dst_name, dst_mode & ~omitted_permissions) != 0)
> {
> error (0, errno, _("setting permissions for %s"),
> quote (dst_name));
> }
> } else {
> error (0, errno, _("cannot create directory %s"),
> quote (dst_name));
> goto un_backup;
> }
This has the same effect as mv does on Solaris 10.
Where the destination directory exists on the new mount point, the
permissions are modified and new files will overwrite existing files in
the destination tree. However, unique files in the destination directory
will not be removed.
The BIG question is:
There is obviously a bug in the GNU mv command as its failing to unlink
the directory, however ...
should the GNU mv command behave the same way as the Solaris mv commnd?
e.g.
f18# ls -la /home/test /test
/home/test:
total 12
drwxr-sr-x. 2 root root 4096 Jul 1 21:18 .
drwxr-xr-x. 14 root root 4096 Jul 1 21:18 ..
-rw-r--r--. 1 root root 2768 Jul 1 21:18 passwd
/test:
total 16
drwxr-xr-x. 2 root root 4096 Jul 1 21:20 .
dr-xr-xr-x. 25 root root 4096 Jul 1 21:09 ..
-rw-r--r--. 1 root root 1162 Jul 1 12:53 group
-rw-r--r--. 1 root root 2777 Jul 1 21:20 passwd
f18# /home/ken/git/fedora/coreutils/src/mv /home/test /
f18# ls -la /home/test /test
ls: cannot access /home/test: No such file or directory
/test:
total 16
drwxr-sr-x. 2 root root 4096 Jul 1 21:18 .
dr-xr-xr-x. 25 root root 4096 Jul 1 21:09 ..
-rw-r--r--. 1 root root 1162 Jul 1 12:53 group
-rw-r--r--. 1 root root 2768 Jul 1 21:18 passwd
or should we really be deleting the directory as defined by the man page
wording "overwrite" and observed by the failed attempt at unlinking ?
e.g.
should we delete /test/group
Which behaviour is correct?
Regards,
Ken Booth
Red Hat UK Ltd
PS: There is a Red Hat bugzilla for this
https://bugzilla.redhat.com/show_bug.cgi?id=980061
--- End Message ---
--- Begin Message ---
Subject: |
Re: bug#14763: mv directory cross-filesystem where destination exists fails to remove dest with EISDIR |
Date: |
Wed, 24 Jul 2013 18:14:29 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20130110 Thunderbird/17.0.2 |
On 07/02/2013 01:29 AM, Ken Booth wrote:
> Hi Eric,
>
> Thanks for the reference to the POSIX standards.
>
> The reason I wrote the patch the way I did was to emulate Solaris behaviour
> for a non-empty directory, however I read at
> http://pubs.opengroup.org/onlinepubs/9699919799/functions/rename.html
> the sentence "If /new/ names an existing directory, it shall be required to
> be an empty directory. "
>
> Therefore I conclude that Solaris is not POSIX compliant.
>
> After reading the instructions you suggested I hope the following patch is in
> the correct format, conforms to the requirements, and is also POSIX compliant
> ...
>
> From 8b6549f321c06ee81262f58c6d7bd7e9c9092b30 Mon Sep 17 00:00:00 2001
> From: Ken Booth <address@hidden>
> Date: Tue, 2 Jul 2013 01:06:32 +0100
> Subject: [PATCH] mv: if inter-device move target is directory use rmdir, not
> unlink
>
> ---
> src/copy.c | 26 ++++++++++++++++++++------
> 1 file changed, 20 insertions(+), 6 deletions(-)
>
> diff --git a/src/copy.c b/src/copy.c
> index c1c8273..5137f27 100644
> --- a/src/copy.c
> +++ b/src/copy.c
> @@ -2173,13 +2173,27 @@ copy_internal (char const *src_name, char const
> *dst_name,
> /* The rename attempt has failed. Remove any existing destination
> file so that a cross-device 'mv' acts as if it were really using
> the rename syscall. */
> - if (unlink (dst_name) != 0 && errno != ENOENT)
> + if (S_ISDIR (src_mode))
> {
> - error (0, errno,
> - _("inter-device move failed: %s to %s; unable to remove
> target"),
> - quote_n (0, src_name), quote_n (1, dst_name));
> - forget_created (src_sb.st_ino, src_sb.st_dev);
> - return false;
> + if (rmdir (dst_name) != 0 && errno != ENOENT)
> + {
> + error (0, errno,
> + _("inter-device move failed: %s to %s; unable to remove
> target directory"),
> + quote_n (0, src_name), quote_n (1, dst_name));
> + forget_created (src_sb.st_ino, src_sb.st_dev);
> + return false;
> + }
> + }
> + else
> + {
> + if (unlink (dst_name) != 0 && errno != ENOENT)
> + {
> + error (0, errno,
> + _("inter-device move failed: %s to %s; unable to remove
> target"),
> + quote_n (0, src_name), quote_n (1, dst_name));
> + forget_created (src_sb.st_ino, src_sb.st_dev);
> + return false;
> + }
> }
>
> new_dst = true;
This looks good, thanks.
Though I don't think there's a need to duplicate the blocks above:
One could minimize to:
- if (unlink (dst_name) != 0 && errno != ENOENT)
+ if ((S_ISDIR (src_mode) ? rmdir : unlink) (dst_name) != 0
+ && errno != ENOENT)
Though I think unlink at least can be a function like macro,
and so the above could cause compile issues on some platforms.
Therefore I'm adjusting to the following equivalent and will then push:
- if (unlink (dst_name) != 0 && errno != ENOENT)
+ if ((S_ISDIR (src_mode) ? rmdir (dst_name) : unlink (dst_name)) != 0
+ && errno != ENOENT)
Note we're checking the type of src which is a bit confusing
since we're removing dst. Now we could check dst as the
overwrite dir with non dir case is checked for earlier (and vice versa).
But I guess this is a double check for this case.
I'll add a comment along these lines.
I'll add a test too.
thanks,
Pádraig.
--- End Message ---