This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Question on internal_get_tmp_var
- From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- To: rth at redhat dot com
- Cc: gcc at gcc dot gnu dot org
- Date: Wed, 23 Jun 04 18:43:45 EDT
- Subject: Re: Question on internal_get_tmp_var
TREE_READONLY for scalar values doesn't help anything. I wouldn't
have expected this to get used for aggregates very often.
It does if you have a SAVE_EXPR for an aggregate. Certainly not common for
large ones, but possible for small ones.
But there's no *harm* in setting it, right?
DECL_INITIAL isn't gimple. Notice that we *clear* DECL_INITIAL
in gimplify_decl_stmt.
I'm running into trouble with the early gimplification of things
because it makes it harder to trace back and find the expression that
a PLACEHOLDER_EXPR is to match. I fixed this particular problem by
making a little deeper in the tree (on a fat pointer instead of a
COMPONENT_REF of it), but I'm a little concerned more that can't be fixed
this easily will come up. Maybe not ...