[linux-cifs-client] Re: cp -p localfile to a winnt, w2k0,
winxp server - and legacy
servers like win98SE and OS/2 - does _not_ succeed. To get that tracked
now, i'll open a samba bugzilla entry, too
Jeff Layton
jlayton at redhat.com
Wed Jul 9 11:02:02 GMT 2008
On Wed, 9 Jul 2008 05:55:34 +0200
Günter Kukkukk <linux at kukkukk.com> wrote:
> I post that here - so it's not going to be forgotten.
>
> Again and again there are posts to many mailing lists,
> that cifs vfs is handling timestamps improperly.
> Regarding legacy servers, special copy operations just fail.
> (the todays smbfs vs. cifs legacy problem)
>
> Test environment:
> - latest git cifs vfs is mounted against
> - winNT
> - WinXP
> - OS/2
> - Win98SE
>
> Unfortunately, this one cannot be simply explained in 3 words ...
> ----------------------------
>
> To make it as clear as possible - the mount commands:
> *****************************************************
> WinNT server:
> =============
> mount -t cifs //wrkgk/test /mnt/winnt -o cred=/root/sec/sec_gk,dir_mode=0775,file_mode=0764,uid=gk,gid=users
>
> WinXP server:
> =============
> mount -t cifs //cboyer/gk /mnt/winxp -o cred=/root/sec/sec_gk,dir_mode=0775,file_mode=0764,uid=gk,gid=users
>
> OS/2 server:
> ============
> mount -t cifs //server01/CIFS /mnt/server -o cred=/root/sec/sec_gk,sec=lanman,servern=SERVER01,port=139,dir_mode=0775,file_mode=0764,uid=gk,gid=users
>
> Win98SE:
> ========
> mount -t cifs //win98/CIFS /mnt/win9x -o cred=/root/sec/sec_gk,servern=WIN98,dir_mode=0775,file_mode=0764,uid=gk,gid=users
>
> ------------------------------------------------------
>
> The results of "cp -p local_file /mnt/some_server":
> ***************************************************
> NOTE:
> cp -p src dst
> the '-p' option should preserve time stamps - at least the
> last modification time!
>
> WinNT:
> ======
> gk at linux700:~/cifstest> stat dummy.tst && cp -p dummy.tst /mnt/winnt/ && sleep 1 && stat /mnt/winnt/dummy.tst
> File: `dummy.tst'
> Size: 51078 Blocks: 104 IO Block: 4096 regular file
> Device: 306h/774d Inode: 1054939 Links: 1
> Access: (0644/-rw-r--r--) Uid: ( 1000/ gk) Gid: ( 100/ users)
> Access: 2008-07-09 04:32:26.000000000 +0200
> Modify: 2007-05-07 21:52:40.000000000 +0200
> Change: 2008-07-09 04:32:24.000000000 +0200
> File: `/mnt/winnt/dummy.tst'
> Size: 51078 Blocks: 104 IO Block: 16384 regular file
> Device: 10h/16d Inode: 11749 Links: 1
> Access: (0764/-rwxrw-r--) Uid: ( 1000/ gk) Gid: ( 100/ users)
> Access: 2008-07-09 04:35:06.854217400 +0200 !!!!!!!!!!!!!!!!!!!!!
> Modify: 2008-07-09 04:35:06.794122600 +0200 !!!!!!!!!!!!!!!!!!!!!
> Change: 2008-07-09 04:35:06.794122600 +0200 !!!!!!!!!!!!!!!!!!!!!
>
> WinXP:
> ======
> gk at linux700:~/cifstest> stat dummy.tst && cp -p dummy.tst /mnt/winxp/ && sleep 1 && stat /mnt/winxp/dummy.tst
> File: `dummy.tst'
> Size: 51078 Blocks: 104 IO Block: 4096 regular file
> Device: 306h/774d Inode: 1054939 Links: 1
> Access: (0644/-rw-r--r--) Uid: ( 1000/ gk) Gid: ( 100/ users)
> Access: 2008-07-09 04:32:24.000000000 +0200 !!!!!!!!!!!!!!!!!!!!!!!!
> Modify: 2007-05-07 21:52:40.000000000 +0200 !!!!!!!!!!!!!!!!!!!!!!!!
> Change: 2008-07-09 04:32:24.000000000 +0200 !!!!!!!!!!!!!!!!!!!!!!!!
> File: `/mnt/winxp/dummy.tst'
> Size: 51078 Blocks: 104 IO Block: 16384 regular file
> Device: 11h/17d Inode: 11484 Links: 1
> Access: (0764/-rwxrw-r--) Uid: ( 1000/ gk) Gid: ( 100/ users)
> Access: 2008-07-09 04:31:31.012852800 +0200
> Modify: 2008-07-09 04:31:31.012852800 +0200
> Change: 2008-07-09 04:31:31.012852800 +0200
>
> OS/2:
> =====
> gk at linux700:~/cifstest> stat dummy.tst && cp -p dummy.tst /mnt/server/ && sleep 1 && stat /mnt/server/dummy.tst
> File: `dummy.tst'
> Size: 51078 Blocks: 104 IO Block: 4096 regular file
> Device: 306h/774d Inode: 1054939 Links: 1
> Access: (0644/-rw-r--r--) Uid: ( 1000/ gk) Gid: ( 100/ users)
> Access: 2008-07-09 04:46:16.000000000 +0200
> Modify: 2007-05-07 21:52:40.000000000 +0200
> Change: 2008-07-09 04:32:24.000000000 +0200
> cp: preserving times for `/mnt/server/dummy.tst': Invalid argument
>
> Win98SE:
> ========
> gk at linux700:~/cifstest> stat dummy.tst && cp -p dummy.tst /mnt/win9x/ && sleep 1 && stat /mnt/win9x/dummy.tst
> File: `dummy.tst'
> Size: 51078 Blocks: 104 IO Block: 4096 regular file
> Device: 306h/774d Inode: 1054939 Links: 1
> Access: (0644/-rw-r--r--) Uid: ( 1000/ gk) Gid: ( 100/ users)
> Access: 2008-07-09 04:36:04.000000000 +0200
> Modify: 2007-05-07 21:52:40.000000000 +0200
> Change: 2008-07-09 04:32:24.000000000 +0200
> cp: preserving times for `/mnt/win9x/dummy.tst': Invalid argument
> -----------------------------------------------------------------
>
> Conclusion:
> ===========
> WinNT and WinXP just _silently_ FAIL!
> OS/2 and Win98SE just fail with an error msg!
>
> All this is "a known cifs failure" for weeks now - Jeff Layton
> has prepared a (large) patch to "cleanup the cifs_setattr()"
> stuff - and I'm waiting to post more legacy functionality.
>
> ... but nothing happens ...
>
> A really sad situation, when realizing that even the samba
> smbfs helpers smbmount, smbmnt and smbumount have now been
> removed in 3.2.x.
>
> Cheers, Günter
Hi Günter,
I think Steve is on holiday this week (it's summer holiday season here
in the US). I think he's waiting for 2.6.26 to be released before
committing more patches. IMO, If you have patches you're just sitting
on, then go ahead and post them and just make it clear that they depend
on other, uncommitted patches. I'm working on some other things too and
I'd like to make sure what I'm doing doesn't conflict with your work
(or anyone else's), and that we're not duplicating each others efforts.
As a side note, it might be nice if Steve had a "devel" branch in his
git tree for stuff intended for the next release so that development
could more easily continue during the mainline -rc phases. Of course, I
need to get to work on a hosted git tree of my own as well ;-)
--
Jeff Layton <jlayton at redhat.com>
More information about the linux-cifs-client
mailing list