This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Graphite middle-end parts review
On Wed, Aug 13, 2008 at 4:07 PM, Richard Guenther <rguenther@suse.de> wrote:
>> No, Jan did the right thing here. I also think that we should have
>> more systematic functions that take care of everything: creating the
>> variable, etc.
>
> Right. Whoever creates the variable should add it. In this
> place you either have created iv without using it in some cases
> or my suggestion is correct. So - I don't agree with you.
Okay, then why isn't add_referenced_var part of create_tmp_var? I
don't like the separation of these two functions, but here is what I
committed to the graphite branch.
2008-08-13 Sebastian Pop <sebastian.pop@amd.com>
* cfgloopmanip.c (create_empty_loop_on_edge): Don't call
add_referenced_var.
* graphite.c (graphite_create_new_loop): Call add_referenced_var.
Index: cfgloopmanip.c
===================================================================
--- cfgloopmanip.c (revision 139073)
+++ cfgloopmanip.c (working copy)
@@ -592,7 +592,6 @@ create_empty_loop_on_edge (edge entry_ed
gsi_commit_edge_inserts ();
}
- add_referenced_var (iv);
standard_iv_increment_position (loop, &gsi, &insert_after);
create_iv (initial_value, stride, iv, loop, &gsi, insert_after,
iv_before, NULL);
Index: graphite.c
===================================================================
--- graphite.c (revision 139047)
+++ graphite.c (working copy)
@@ -3140,6 +3140,8 @@ graphite_create_new_loop (scop_p scop, e
SCOP_PARAMS (scop),
ivstack);
ivvar = create_tmp_var (integer_type_node, "graphiteIV");
+ add_referenced_var (ivvar);
+
upb = clast_to_gcc_expression (stmt->UB, integer_type_node,
SCOP_PARAMS (scop),
ivstack);