This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: GIMPLE / induction variable analysis
- From: Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>
- To: "Jay L. T. Cornwall" <jay dot cornwall at imperial dot ac dot uk>
- Cc: gcc at gcc dot gnu dot org
- Date: Sat, 10 Jun 2006 10:25:38 +0200
- Subject: Re: GIMPLE / induction variable analysis
- References: <4489D582.5040901@imperial.ac.uk>
Hello,
> I'm looking for an induction variable analysis which will operate at the
> GIMPLE level. So far, I've found...
>
> loop-iv.c -- RTL induction variable analysis
> tree-ssa-loop-ivopts.c -- GIMPLE induction variable analysis
no, tree-level induction variable analysis is in
tree-scalar-evolution.c. What you are looking for most likely is
simple_iv function (or if you are adventurous and want to handle
non-affine induction variables or evolutions in several loops at once
as well, analyze_scalar_evolution and instantiate_parameters functions).
Zdenek
> The latter appears to be the right choice but nearly all of the
> functions - and certainly the key data structures - are placed in the .c
> file, hidden away from the outside world.
>
> Is there a reason that this interface hasn't been exposed in a header
> file? I can certainly duplicate the interface into a header - for
> private development purposes - but if there's a good reason not do this,
> is there a better approach?
>
> Thanks,
>
> --
> Jay L.T. Cornwall, http://www.esuna.co.uk/~jay/
> PhD Student
> Imperial College London