This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-profiling/RFC] Extending passmanager to deal with IPA passes
- From: Andrew Pinski <pinskia at physics dot uc dot edu>
- To: Jan Hubicka <jh at suse dot cz>
- Cc: gcc-patches at gcc dot gnu dot org, rth at redhat dot com
- Date: Tue, 9 Nov 2004 20:58:32 -0500
- Subject: Re: [tree-profiling/RFC] Extending passmanager to deal with IPA passes
- References: <20041025091302.GN21413@kam.mff.cuni.cz>
On Oct 25, 2004, at 5:13 AM, Jan Hubicka wrote:
Hi,
this patch extends the passmanager to deal with IPA passes. This is
done by adding 4 new hooks neccesary to the PM structure and adding
some
extra entry points to do the IPA analysis and optimizations to the
tree-optimize.
In general I would like to replace the tree_rest_of_compilation by just
set of nested passes so the passmanager is completely driven by
cgraphunit and all the other actions are driven by the PM.
I was look into using the pass manager for an IPA pass when I noticed
that we do the following order for passes:
function analysis
execute
variable analysis
Which seems wrong as we should do the following order instead:
function analysis
variable analysis
execute
As we generate some information by the two analysis functions which
should be used by the execute function which do some more analysis
of the data which is used by the modify functions.
I think the order right now causes the static-var pass to generate
not correct information as of right now.
Thanks,
Andrew Pinski