ppc64le regressions due to recent change

[Marc Glisse]

On Thu, 3 Oct 2019, Jan Hubicka wrote:

>>> Tests that now fail, but worked before (15 tests):
>>>
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++14  scan-tree-dump-times fre3 ";; Function" 1
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++14  scan-tree-dump-times fre3 "free" 14
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++14  scan-tree-dump-times optimized "free" 0
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++17  scan-tree-dump-times fre3 ";; Function" 1
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++17  scan-tree-dump-times fre3 "free" 14
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++17  scan-tree-dump-times optimized "free" 0
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++98  scan-tree-dump-times fre3 ";; Function" 1
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++98  scan-tree-dump-times fre3 "free" 14
>>> g++.dg/tree-ssa/pr61034.C  -std=gnu++98  scan-tree-dump-times optimized "free" 0
>>
>> There were others, but starting with these seems as good as any.
>
> Sorry for that. Seems I managed to leave this one out of the final
> patch. It is very sensitive to how things are inlined. If it breaks
> again I guess I will try to work out explicit noinline/always_inline
> attributes to make it to work.

If you want to cheat and consider it as a test only for FRE / DCE, 
__attribute__((flatten)) on f should work. But it can be useful to keep it 
in a state where it still tests the inliner as well...

-- 
Marc Glisse