On Mar 3, 2021, Segher Boessenkool <seg...@kernel.crashing.org> wrote:
> On Wed, Mar 03, 2021 at 04:45:23PM -0300, Alexandre Oliva wrote: >> On Mar 2, 2021, Segher Boessenkool <seg...@kernel.crashing.org> wrote: >> >> > This is PR99352 now. >> >> Thanks. I've filed PR99371 for the add_options_for_sqrt_insn >> incompleteness, > As I said before, I don't agree with that. Maybe you'll change of mind if you try to make sense of why the proc has, since its inception, added vfp options to enable sqrt on arm, regardless of whether vfp is available with the processor being tested, and realize this is not different from the case of powerpc. > If a user disabled it, we should *not* reenable it, that reduces > testing surface. It's skipping the test, as the change you propose, that reduces testing surface, when testing only a configuration that ends up skipping it. Now, if you're testing multiple combinations, skipping or running does *not* change the test surface. So rejecting Eric's patch makes for a no-op in one case, and a reduction in another. Whereas installing it makes for a no-op in one case, and an increase in another. Please explain how you came to the conclusion that this amounts to reducing hte test surface. Something appears to be amiss in that reasoning. >> and PR99372 for the gimplefe-28.c ICE. > This is fixed trivially by the PR99352 patch as far as I can see. If your patch also deals with the ICE that appears with the options named in PR99372, great. > Please verify (I'll post it later today). Please Cc me so I don't miss it, thanks, -- Alexandre Oliva, happy hacker https://FSFLA.org/blogs/lxo/ Free Software Activist GNU Toolchain Engineer Vim, Vi, Voltei pro Emacs -- GNUlius Caesar
