Comment 42 for bug 1238446

It looks like parted's gpt code simply throws out the upper 8 bits of each 16 bit character, which corrupts non ascii names. Later it tries to treat this corrupted name as a multi byte string ( which I think means UTF-8 ) and convert it to a wide character string, and this fails.

I'm not sure how to convert UTF-16 to the current locale encoding.