Hello, when conducting some automated filesystem tests, I've discovered
that some filesystems are unable to cope with non-BMP characters. While
most usual characters are in BMP, I still think that supporting then is
important. Perhaps, the single most important reason to support them is
that they come up in some personal names. The series has 2 patches to
prepare the ground by changing NLS functions uni2char and char2uni to
work on full unicode_t and another patch for UTF-16 conversion functions.
Remaining 6 patches are for specific filesystems.The filesystems I haven't
fixed are ncpfs and cifs since I don't have the relevant test environment
but it should be easy for someone with one to extend it for them as well.
I've used mainly 2 following characters for test:
茝 Some rare CJK symbol, no idea what it means, picked it up at random.
😁 An emoticon.
I send this explanation and first 2 patches to all concerned parties. Rest
is sent to LKML, linux-fsdevel and relevant filesystem maintaintainer/list only
Vladimir 'φ-coder/phcoder' Serbinenko