2.25 freeze status

[Joseph Myers]

On Tue, 31 Jan 2017, Carlos O'Donell wrote:

> I ran out of time today doing a semi-final round of testing on aarch64, 
> armv7hl, ppc (still building), ppc64, ppc64le, s390x, s390 (still testing),
> x86_64, and x86 (wiki updated).

For reference regarding some of the failures you listed on the wiki:

* The tgamma failures for AArch64 with an old compiler indicate you are 
missing the GCC fix for 
<https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66731>.  (The other math/ 
failures there are probably compiler bugs but I don't have specific 
references.)

* The test-ldouble failures for x86 should just have "make regen-ulps" run 
on the affected system (the functions in question use x87 transcendental 
math instructions so the exact results are expected to depend on the 
hardware).

> I prioritized making sure we had no surprises there over this final patch.
> 
> > - Fix for bug 20019 causes linker errors for shared libraries using
> >   longjmp
> 
> Looking to get consensus with H.J. over this issue. I would hope we reach
> a conclusion tomorrow.

I think we should revert the problem change / changes to get 2.25 to the 
same state as 2.24.

-- 
Joseph S. Myers
joseph@codesourcery.com