page01 and page1e in nls_iso8859-14.c are identical, so the U+01xx
table is incorrectly replaced by the U+1Exx table. Both are used by
page_uni2charset, causing uni2char() to map 29 code points to the wrong
bytes.
charset2uni() is unaffected and matches the Unicode 8859-14 mapping.
For example:
U+010A -> 0xa6 -> U+1E0A
U+010B -> 0xab -> U+1E0B
U+1E20 -> 0xb2 -> U+0120
22 code points that are not part of ISO 8859-14 are also incorrectly
accepted instead of being rejected.
Regenerate page01 and page1e from charset2uni. glibc iconv agrees with
the resulting mappings, and a full BMP round trip is now lossless.
Signed-off-by: Ali Ahmet Memiş <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Christian Brauner (Amutable) <[email protected]>