This is the mail archive of the
mailing list for the GCC project.
Re: [PR58315] reset inlined debug vars at return-to point
- From: Alexandre Oliva <aoliva at redhat dot com>
- To: Richard Biener <richard dot guenther at gmail dot com>
- Cc: Jakub Jelinek <jakub at redhat dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Thu, 26 Feb 2015 13:42:01 -0300
- Subject: Re: [PR58315] reset inlined debug vars at return-to point
- Authentication-results: sourceware.org; auth=none
- References: <orlhjm2usl dot fsf at livre dot home> <CAFiYyc1a+cMHKk-dAJ3SbErhDotEGFRekK=xBy+DbHpFpZWE2g at mail dot gmail dot com> <20150225161256 dot GT1746 at tucnak dot redhat dot com> <or7fv53d3m dot fsf at livre dot home> <20150225212231 dot GX1746 at tucnak dot redhat dot com> <ortwy91qyi dot fsf at livre dot home> <CAFiYyc16cJOfeS=TyurCx_x50keHergEgxGT4AcGa9owQnHK-Q at mail dot gmail dot com>
On Feb 26, 2015, Richard Biener <firstname.lastname@example.org> wrote:
> and more concentrating on the effect of your patch as opposed to
> debug stmt philosophy.
> (which looks reasonable minus code-motion issues).
> (but we might still want to compute it during var-tracking
> if at a later PC range the scope will be active again).
By taking debug stmt philosophy out of your reasoning, you come to a
conclusion that directly contradicts the very philosophy. I.e., you're
not setting the philosophy aside, you're setting yourself up for failure
within that philosophy.
Your note on code-motion issues shows another weakness, not in the
philosophy I endorse, but on rather on the one you do. If annotations
don't move, and they carry all the information debuggers need to care
about, there's no code-motion issue whatsoever, and moving executable
instructions would not have any effects whatsoever. Sure, we're not
there yet, but if we want to get there, taking steps in the opposite
direction won't take us there any time soon.
> Thus consider the suggestion to insert reset debug insns at the beginning
> of var-tracking and at points where a scope becomes dead (thus after
> points where in a backward CFG walk a scope becomes live).
The point where the scope becomes dead under what philosophical line?
I believe we've already established that trying to recover that
information after throwing it away is cheaper but error prone.
I believe we've already agreed that we don't want debug information to
Alexandre Oliva, freedom fighter http://FSFLA.org/~lxoliva/
You must be the change you wish to see in the world. -- Gandhi
Be Free! -- http://FSFLA.org/ FSF Latin America board member
Free Software Evangelist|Red Hat Brasil GNU Toolchain Engineer