[Bug ipa/65237] [5 Regression] r221040 caused many regressions

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65237

--- Comment #6 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Martin Liška from comment #3)
> (In reply to howarth from comment #2)
> > (In reply to Iain Sandoe from comment #1)
> > 
> > > 
> > > On Darwin we also see:
> > > gcc.dg/attr-noinline.c (all cases fail)
> > 
> > Didin't Martin analyze this failure as the test case just needing "/* {
> > dg-require-alias "" }  */"?
> > 
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65150#c13
> 
> Yes, I think the test should be decoared with dg-require-alias.

hmm.. so should we understand that we used to pass that test .. "incorrectly?"
.. 
.. it's not clear to me that changing the test to require symbol alias isn't
still a regression … (I'd like to understand that the code-gen is equivalent).