[v3] c++0x <system_error> CD updates

[DJ Delorie]

> I see your point, on the other hand seems to me a bit "unfair"
> pointing out that in a message following up to Chris' patch, which
> didn't touch that line.

Sorry, it only just broke, and he just touched that file, that's as
far as I checked.

> Also, we already have something very similar in, eg, ios_base.h, and
> nobdy complained to date, it's *very* old (and critical) code.

libstdc++ built yesterday morning, so either m32c skips that code, or
it's different somehow.