mbtowc(3): set errno to EILSEQ if an incomplete character is passed.

[pfg]

Summary

According to POSIX, The mbtowc() function shall fail if:
[EILSEQ] An invalid character sequence is detected.

Obtained from: OpenBSD (Ingo Schwarze)

Test Plan

The related test from kyua consistently shows an expected failure.
...
20160301-031439 I 58919 utils/cmdline/ui.cpp:103: stdout: lib/libc/locale/mbtowc_test:mbtowc ->
20160301-031439 I 58919 engine/scheduler.cpp:1084: Spawning /usr/tests/lib/libc/locale/mbtowc_test:mbtowc
20160301-031439 I 58919 utils/process/executor.cpp:774: Spawned subprocess with exec_handle 87696
20160301-031439 I 58919 utils/process/executor.cpp:630: Waited for subprocess with exec_handle 87696
20160301-031439 I 58919 utils/process/executor.cpp:368: Cleaning up exit_handle for exec_handle 87696
20160301-031439 I 58919 utils/cmdline/ui.cpp:103: stdout: expected_failure: zh_CN.GB18030: /usr/src/contrib/netbsd-tests/lib/libc/locale/t_mbtowc.c:122: ret != (size_t)-1 not met [0.004s]
...

[bapt]

Looks good to me have you checked that it does not break the test suite? I remember that some tests were failing after I did some of those changes (and I needed to fix those tests)

[pfg]

In D5496#117148, @bapt wrote:

Looks good to me have you checked that it does not break the test suite? I remember that some tests were failing after I did some of those changes (and I needed to fix those tests)

Interesting. Thanks for the feedback.
It would be really surprising if this would cause an error on the testsuite .. will check :).

[pfg]

Committed as r296278.