[RFC] atomics vs. uniprocessor builds

[Joseph Myers]

On Tue, 25 Nov 2014, Torvald Riegel wrote:

> My guess would be that uniprocessor builds are rarely used, so it might
> be fine to just drop the UP optimization on the atomics.  Keeping the

Yes.  More generally: conditionals that are not active for any in-tree 
glibc configuration (possibly together with documented configure options), 
and are not in code shared with other projects such as gnulib, libgcc, 
gettext, GMP, should be presumed dead and can be removed in the absence of 
concrete reasons to keep them (e.g. if the conditional serves a useful 
purpose as documentation of what the requirements of the code are, where a 
new port might plausibly need to implement something new - a conditional 
that causes a build failure for a system with 128-bit intmax_t might make 
sense to keep, one that verifies int is at least 32-bit wouldn't outside 
of shared code).

So I think all UP conditionals should be removed.

-- 
Joseph S. Myers
joseph@codesourcery.com