On Fri, Mar 13, 2020 at 07:30:17AM +0100, Richard Biener wrote:
> On March 12, 2020 11:29:45 PM GMT+01:00, Segher Boessenkool
> <seg...@kernel.crashing.org> wrote:
> >The df dataflow solvers use the aux field in the basic_block struct,
> >although that is reserved for any use by passes. And not only that,
> >it is required that you set all such fields to NULL before calling
> >the solvers, or you quietly get wrong results.
> >
> >This changes the solvers to use a local array for last_change_age
> >instead, just like it already had a local array for last_visit_age.
> >
> >Tested on powerpc64-linux {-m32,-m64}. Also tested with the tests for
> >PR94042, which it also solves (I need to confirm that still though,
> >there are other testsuite problems interfering with my testing).
>
> OK for trunk and affected branches (can you report back how old this issue
> is?)
It's from r161197:
commit 1a0f3fa13745c4052c53e6d84c64539fb5f57e00
Author: Jan Hubicka <j...@suse.cz>
Date: Tue Jun 22 17:51:15 2010 +0200
This is https://patchwork.ozlabs.org/patch/55394/ btw, which has the ML
discussion and everything.
$ git branch -r --contains 1a0f3fa13745|grep fsf.releases
fsf/releases/gcc-4.6
fsf/releases/gcc-4.7
fsf/releases/gcc-4.8
fsf/releases/gcc-4.9
fsf/releases/gcc-5
fsf/releases/gcc-6
fsf/releases/gcc-7
fsf/releases/gcc-8
fsf/releases/gcc-9
(Maybe there was a backport to 4.5, but that is far back enough already ;-) )
Segher
