generalize errno handling

[Mike Frysinger]

On Wed, Mar 23, 2011 at 5:57 PM, Peter Mazinger wrote:
>> >> the point is to not have the overhead of an indirect pointer in the
>> >> non-threaded case.  are you suggesting we force that overhead when it
>> >> is unnecessary ?
>> >
>> > yes
>>
>> then i'd say find a different way.  it cant be that hard to hide the
>> indirection in a header with CPP.
>>
> now that I look closer, the
> #define h_errno (*__h_errno_location ())
> is already unguarded by THREADS since a commit for NPTL, see below

then it's a bug that needs fixing, not further enshrining
-mike