Hi, On Fri, Sep 16, 2011 at 04:29:44PM +0200, Jan Hubicka wrote: > > > > * ipa-prop.h (jump_func_type): Updated comments. > > (ipa_known_type_data): New type. > > (ipa_jump_func): Use it to describe known type jump functions. > > * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Updated to > > reflect the new known type jump function contents. > > (compute_known_type_jump_func): Likewise. > > (combine_known_type_and_ancestor_jfs): Likewise. > > (try_make_edge_direct_virtual_call): Likewise. > > (ipa_write_jump_function): Likewise. > > (ipa_read_jump_function): Likewise. > > * ipa-cp.c (ipa_value_from_known_type_jfunc): New function. > > (ipa_value_from_jfunc): Use ipa_value_from_known_type_jfunc. > > (propagate_accross_jump_function): Likewise. > > OK. If we saved just one pointer to the actual type (i.e. not > BINFO), would it be any sanier?
I'm not sure if I understand the question. In any event, the base_type is certainly not enough, the offset is 100% necessary - it identifies the vptr in the structure which we have to know because the indices (tokens) in OBJ_TYPE_REFs are relative to what that vptr, not the first vptr, points to. I have attempted to remove the component_type (which means to remove the expected_type in get_binfo_at_offset) but my first simple attempts could lead to missed optimizations in cases we can currently handle. I still have one idea that might work, I will try. Meanwhile, I will re-test the patch on a current svn checkout and commit it if all goes fine. Thanks, Martin
