samba - Oct 2005 - unicode/control characters displaying as ? with cifsfs

greeting samba.general;

	i am trying to set up a backup server to archive files from
	a windows 2000 fileserver, and i can't seem to get filenames
	containing non-ascii characters to appear correctly on the
	linux box.

	my windows server:
	windows 2000 service pack 4

	my linux server:
	debian 3.1
	kernel 2.6.13.3 with these options:
	CONFIG_CIFS=m
	# CONFIG_CIFS_STATS is not set
	# CONFIG_CIFS_XATTR is not set
	# CONFIG_CIFS_EXPERIMENTAL is not set
	samba 3.0.14a-3

	my mount command looks like this:
mount -t cifs //YCS1610-767/Jobs /mnt/prinergy/jobs/ -o
username=imageserver,password=xxx

	listing a typical directory:
admin@mpi-images:~$ ls "/mnt/prinergy/jobs/ System 5 Bearer Bars/"
57686e_f.srd  Icon?  System 5 bottom bearer.ai6  System 5 top bearer.ai6

	when i try to copy the "Icon?" file, it will fail with:
admin@mpi-images:~$ cp "/mnt/prinergy/jobs/ System 5 Bearer
Bars/Icon?" /tmp
cp: cannot stat `/mnt/prinergy/jobs/ System 5 Bearer Bars/Icon?': No such
file or directory

	obviously this is a problem. according to what i can find online,
	cifsfs is supposed to handle unicode correctly, yet it doesn't seem
	to be working. fwiw the same mount command using "mount -f smbfs"
	exhibits the same behaviour.

	any suggestions?

	tia
	henry

henry j. mason wrote:
>     greeting samba.general;
> 
>     i am trying to set up a backup server to archive files from
>     a windows 2000 fileserver, and i can't seem to get filenames
>     containing non-ascii characters to appear correctly on the
>     linux box.
	after some more digging on the web, i found that i need to
	generate a UTF-8 character set encoding. by default, debian
	only comes with a bare minumum, and the default charset
	happens to be en_US.iso88591 - which doesn't match the UTF-8
	encoding used by windows. so, to do this under debian (as root):

mpi-images:~# dpkg-reconfigure locales

	scroll down until you see the UTF-8 encoding for your locale,
	in this case, en_US.UTF-8. select the encoding, and choose
	'Ok'. debian will generate the locales for you. check their
	existence with:

mpi-images:~# locale -a
C
POSIX
en_US
en_US.iso88591
en_US.utf8

	then, the magic mount command is:
mpi-images:~# mount -t cifs //YCS1610-767/Jobs /mnt/prinergy/jobs/ -o
username=imageserver,password=xxx,iocharset=utf8

	now, when i list that same directory:
admin@mpi-images:~$ ls -l "/mnt/prinergy/jobs/ System 5 Bearer Bars/"
total 592
-rwxrwSrwt  1 root root 537856 2004-01-05 10:24 57686e_f.srd
-rwxrwSrwt  1 root root      0 2003-06-19 13:14 Icon?
-rwxrwSrwt  1 root root  32455 2003-07-09 14:05 System 5 bottom bearer.ai6
-rwxrwSrwt  1 root root  32242 2003-07-09 14:06 System 5 top bearer.ai6

	the stupid control character shows up correctly as whatever
	it is. and the copy command executes happily:

admin@mpi-images:~$ cp "/mnt/prinergy/jobs/ System 5 Bearer
Bars/Icon?" /tmp/
admin@mpi-images:~$

	voila!
	hope this helps someone else :)

	regards,
	henry