[Bug locale/19519] iconv(1) with -c option hangs on illegal multi-byte sequences (CVE-2016-10228)

["arjun.is at lostca dot se" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=19519

--- Comment #12 from Arjun Shankar <arjun.is at lostca dot se> ---
I quickly want to point out that passing an input character that requires
transliteration and then passing //translit as the second suffix also leads to
erroneous program behaviour (it reports an error instead of performing
transliteration). Some test cases:

Exits successfully with 'A' written to stdout (because //translit is the only
or first suffix):

$ echo -en '\xc3\x81' | iconv -f utf8 -t "us-ascii//translit"
$ echo -en '\xc3\x81' | iconv -f utf8 -t "us-ascii//translit//ignore"

No transliteration and corresponding "illegal input sequence" error thrown,
even though //translit is passed, but it is incorrectly ignored by the program
because it is not the first suffix:

$ echo -en '\xc3\x81' | iconv -f utf8 -t "us-ascii//ignore//translit"

Exits without error messages but (incorrectly) writes nothing to stdout. This
is because 'translit' got ignored (2nd suffix), but '-c' leads to silently
dropping the input character that wasn't transliterated:

$ echo -en '\xc3\x81' | iconv -f utf8 -t "us-ascii//ignore//translit" -c

-- 
You are receiving this mail because:
You are on the CC list for the bug.