System: Red Hat 9, kernel 2.4.20-24.9 (custom build with POSIX ACL support), glibc 2.3.2-11.9. Samba 3.0.1 (built from official source RPM, --with-acl-support added) smb.conf: [global] dos charset = IBM866 unix charset = KOI8-R Windows 2000 or XP client: Locale: Russian Language settings: Cyrillic (default), Western Europe and United States Input languages: English (United states) (default), Russian Create a file named "foo №1". The funny character is NUMERO SIGN, 0x2116 in Unicode, 0xfc in IBM866 encoding, and not encoded at all in KOI8-R. File gets created successfully, but its name changes to "foo\ ^V\!1" (as seen by zsh.exe from Cygwin) -- that's raw unicode embedded into ASCII string!
Where do you create the file? On Unix or on Windows?
On Windows.
This is probably a duplicate of Bug 1578 and the patch from there should help. Please test and report the results.
The problem is indeed fixed (tested on 3.0.7).
sorry for the same, cleaning up the database to prevent unecessary reopens of bugs.