[PATCH] Reduce PHI node by 4/8 bytes

[Andrew_Pinski at PlayStation dot Sony dot Com]

During the tuple decussion, I mentioned that PHIs' memory usage could be 
reduced by remove the type field.  This patch implements this suggestion 
by using the work that was done for GIMPLE_MODIFY_STMT.  A couple changes 
to the middle-end more than just changing the base object type was needed 
because some passes uses TREE_CHAIN of the phi node when they should have 
been using PHI_CHAIN.  Also some of the front-ends needed to be changed 
for their definition of lang_tree_node for the chain_next GTY.  This patch 
adds a new macro called GENERIC_NEXT as recommended by Deigo to simplify 
that code.

OK? Bootstrapped and tested on powerpc64-linux-gnu with no regressions, 
also cross built to spu-elf and tested with no regressions.

Thanks,
Andrew Pinski

ChangeLog:

        * tree.h (GIMPLE_TUPLE_P): Also true for PHI_NODEs.
        (GENERIC_NEXT): New function macro.
        (PHI_CHAIN): Use phi_node's new chain variable.
        (tree_phi_node): Change tree_common to tree_base
        and add the chain field.
        * tree-phinodes.c (make_phi_node): Don't set
        TREE_TYPE on the new node.
        * c-decl.c (lang_tree_node): Use GENERIC_NEXT
        instead of checking GIMPLE_TUPLE_P in chain_next.
        * tree-vect-transform.c
        (get_initial_def_for_induction): Look at
        PHI_RESULT_TREE for the type of the phi node.
        (update_vuses_to_preheader): Use PHI_CHAIN
        instead of TREE_CHAIN on the phi node.
        * tree-ssa-structalias.c (compute_points_to_sets):
        Likewise.
        (ipa_pta_execute): Likewise.

cp/ChangeLog:
        * cp-tree.h (lang_tree_node): Use GENERIC_NEXT
        instead of checking GIMPLE_STMT_P in chain_next.

ada/ChangeLog:
        * ada-tree.h (lang_tree_node): Use GENERIC_NEXT
        instead of checking GIMPLE_STMT_P in chain_next.

java/ChangeLog:
        * java-tree.h (lang_tree_node): Use GENERIC_NEXT
        instead of checking GIMPLE_STMT_P in chain_next.

fortran/ChangeLog:
        * f95-lang.c (lang_tree_node): Use GENERIC_NEXT
        instead of checking GIMPLE_STMT_P in chain_next.

Attachment: phinode.diff.txt
Description: Text document