This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Extend tree-call-cdce to calls whose result is used
- From: Michael Matz <matz at suse dot de>
- To: Richard Biener <richard dot guenther at gmail dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>, richard dot sandiford at arm dot com
- Date: Mon, 16 Nov 2015 17:19:34 +0100 (CET)
- Subject: Re: Extend tree-call-cdce to calls whose result is used
- Authentication-results: sourceware.org; auth=none
- References: <874mgyorwm dot fsf at e105548-lin dot cambridge dot arm dot com> <alpine dot LSU dot 2 dot 20 dot 1511091509260 dot 11029 at wotan dot suse dot de> <87r3jznqvg dot fsf at e105548-lin dot cambridge dot arm dot com> <alpine dot LSU dot 2 dot 20 dot 1511091652400 dot 11029 at wotan dot suse dot de> <87oaf3m4p3 dot fsf at e105548-lin dot cambridge dot arm dot com> <alpine dot LSU dot 2 dot 20 dot 1511092122060 dot 11029 at wotan dot suse dot de> <CAFiYyc1A2hbvFa0OYmLzRRmDbzaB6+o9iRfdWX_POW-aHJF1DA at mail dot gmail dot com> <87lha2kpqt dot fsf at e105548-lin dot cambridge dot arm dot com> <CAFiYyc1N6WAPj1KnxpRXjd55Y-VgQ8xtfM2qiK-9rxFK8gOC5Q at mail dot gmail dot com>
Hi,
On Mon, 16 Nov 2015, Richard Biener wrote:
> >> Which would leave us with a lowering stage early in the main
> >> optimization pipeline - I think fold_builtins pass is way too late
> >> but any "folding" pass will do (like forwprop or backprop where the
> >> latter might be better because it might end up computing FP "ranges"
> >> to improve the initial lowering code).
> >
> > This isn't at all related to what backprop is doing though. backprop
> > is about optimising definitions based on information about all uses.
Right, I think backprop would be even worse than call_cdce, that pass has
a completely different structure.
> >> Of course call_cdce is as good as long as it still exists.
> >
> > Does this meann that you're not against the patch in principle (i.e.
> > keeping call_cdce for now and extending it in the way that this patch
> > does)?
>
> Yes, I'm fine with extending call_cdce. Of course I'd happily approve a
> patch dissolving it into somewhere where it makes more sense. But this
> shouldn't block this patch.
Okay, I like merging passes, so I'll try to do that, once the stuff is in
:)
Ciao,
Michael.