I have a Debian/testing server with samba 3.0.5 running. A Person from a Windows Box writes a file with a japense filename to the samba server. The file is correctly encoded in UTF-8. I log in with a gnome-terminal, set the LANG to en_US.UTF-8 and I am able to view the filename correctly. Now I log in from my linux box (gentoo x86 stable with samba 3.0.5) with smbclient I can see thet file too correctly, but if I now mount that share with smbmount (I used various settings for iocharset and codepage: utf8, utf-8, etc or none) the file is not shown. On the server I have log entries that say that iconv is used to convert that filename to utf-8, but I never get any readable output. If I look into that dir via NFS, I can see the file correctly (NFS mounted with 2.6.5-gentoo-1 to NFS server on the same debian/testing box with 2.6.4 vanilla). Somehow there are some bugs with UTF-8 files and the encoding of non ASCII characters.
I have no problems with some umlaut characters and utf8 in smbfs. smbfs however is not part of samba and it is not very well maintained. With recent 2.6.x kernels better use cifs, it is more stable and well maintained.
okay, I will give CIFS a try and check if I can see the japanese characters then.