Re: Cgraph alias reorg 15/14 (New infrastructure for same body aliases)

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Sat, Jun 11, 2011 at 9:15 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Sat, Jun 11, 2011 at 5:55 AM, Jan Hubicka <hubicka@ucw.cz> wrote:
>> Hi,
>> this patch complettes the same body alias rework by removing the old same body
>> alias code and adding new representation. ?Same body aliases are now separate
>> function nodes that have IPA_REF_ALIAS reference to the node they are alias of.
>>
>> I am still getting one failure:
>> FAIL: g++.dg/torture/pr43879-1_1.C
>>
>> It tests IPA PTA in presence of same body aliases. ?I honestly have no idea
>> what is wrong there. ?I decided to go ahead with the patch anyway, given the
>> current state of affair of aliases and IPA-PTA. Hope Richard will help me
>> fixing this on Monday.
>>
>> The major pain with this change is that C++ FE gets same body aliases wrong.
>> It creates them with random visibilities that are not final ones, it also
>> creates them late in game via DECL_ASSEMBLER_NAME langhook, it forgets to set
>> them same_comdat_group links and it do not set DECLARED_INLINE and
>> DISREGARD_INLINE_LIMITS flags.
>>
>> I fix that up at cgraphunit side for now, but it would be cool to resolve those
>> problems.
>>
>> I will followup with two extra fixes needed to fully build Mozilla with LTO
>> again. ?Once things settle down, I will switch real aliases to the new
>> infrastructure, too.
>>
>> Bootstrapped/regtested x86_64-linux, will commit it today.
>>
>>
>> ? ? ? ?* lto-symtab.c (lto_cgraph_replace_node): Kill same body alias code.
>> ? ? ? ?(lto_symtab_resolve_can_prevail_p): Likewise.
>> ? ? ? ?(lto_symtab_merge_cgraph_nodes): Update merging of aliases.
>> ? ? ? ?* cgraph.c (same_body_aliases_done): New global var.
>> ? ? ? ?(cgraph_same_body_alias_1): Rename to ...
>> ? ? ? ?(cgraph_create_function_alias): ... this one; reorg to new
>> ? ? ? ?representation.
>> ? ? ? ?(cgraph_same_body_alias): Use cgraph_create_function_alias;
>> ? ? ? ?record references when asked to.
>> ? ? ? ?(cgraph_add_thunk): Fix formating.
>> ? ? ? ?(cgraph_get_node): Kill same body alias code.
>> ? ? ? ?(cgraph_node_for_asm): Likewise.
>> ? ? ? ?(cgraph_remove_same_body_alias): Remove.
>> ? ? ? ?(cgraph_remove_node): Kill same body alias code.
>> ? ? ? ?(cgraph_mark_address_taken_node): Mark also the aliased function
>> ? ? ? ?as having address taken.
>> ? ? ? ?(dump_cgraph_node): Dump same body aliases.
>> ? ? ? ?(cgraph_for_node_thunks_and_aliases): Update for new alias
>> ? ? ? ?representation.
>> ? ? ? ?(cgraph_for_node_and_aliases): Likewise.
>> ? ? ? ?* cgraph.h (same_body): Kll pointer.
>> ? ? ? ?(same_body_alias): Update comment.
>> ? ? ? ?(same_body_aliases_done): Declare.
>> ? ? ? ?(cgraph_remove_same_body_alias): Remove declaration.
>> ? ? ? ?(cgraph_create_function_alias): Declare.
>> ? ? ? ?(cgraph_process_same_body_aliases): Declare.
>> ? ? ? ?(cgraph_function_with_gimple_body_p): Check for alias.
>> ? ? ? ?(cgraph_can_remove_if_no_direct_calls_p): Look for aliases.
>> ? ? ? ?(cgraph_alias_aliased_node): New function.
>> ? ? ? ?(cgraph_function_node): Update for new aliases.
>> ? ? ? ?(cgraph_function_or_thunk_node): Likewise.
>> ? ? ? ?* ipa-inline-transform.c (can_remove_node_now_p): Look for aliases.
>> ? ? ? ?(inline_call): Remove dead aliases.
>> ? ? ? ?* cgraphunit.c (cgraph_decide_is_function_needed): Disable assembler name
>> ? ? ? ?hack for same body aliases.
>> ? ? ? ?(clone_of_p): Look through aliases.
>> ? ? ? ?(verify_cgraph_node): Verify aliases.
>> ? ? ? ?(cgraph_analyze_function): Analyze aliases; fixup C++ bugs.
>> ? ? ? ?(cgraph_process_same_body_aliases): New function.
>> ? ? ? ?(process_function_and_variable_attributes): Disable weakref warning on
>> ? ? ? ?alias.
>> ? ? ? ?(cgraph_analyze_functions): Handle aliases.
>> ? ? ? ?(cgraph_mark_functions_to_output): Handle aliases same way as thunks.
>> ? ? ? ?(assemble_thunks): Rename to ...
>> ? ? ? ?(assemble_thunks_and_aliases): ... this one; handle aliases, too.
>> ? ? ? ?(cgraph_expand_function): Remove alias output code.
>> ? ? ? ?(cgraph_output_in_order): Skip aliases.
>> ? ? ? ?(cgraph_preserve_function_body_p): Aliases don't need preserving.
>> ? ? ? ?* ipa-ref.c (ipa_ref_use_name): Add alias reference.
>> ? ? ? ?(ipa_record_reference): Do not assert on alias references.
>> ? ? ? ?(ipa_ref_has_aliases_p): New function.
>> ? ? ? ?* ipa-ref.h (enum ipa_ref_use): Add IPA_REF_ALIAS.
>> ? ? ? ?(ipa_ref_has_aliases_p): Declare.
>> ? ? ? ?* lto-cgraph.c (lto_output_node): Handle aliases.
>> ? ? ? ?(input_node): Likewise.
>> ? ? ? ?* lto-streamer-out.c (lto_output): Skip aliases.
>> ? ? ? ?(produce_symtab): Kill same_body_alias code.
>> ? ? ? ?* ipa-utils.c (ipa_reverse_postorder): Add FIXME.
>> ? ? ? ?(ipa_reverse_postorder): Use cgraph_only_called_directly_or_aliased_p.
>> ? ? ? ?* ipa-inline.c (update_caller_keys): Walk aliases.
>> ? ? ? ?(inline_small_functions): Fix thinko in previous patch.
>> ? ? ? ?* ipa.c (cgraph_externally_visible_p): Do not walk aliases.
>> ? ? ? ?(function_and_variable_visibility): Do not walk same body aliases.
>> ? ? ? ?* tree-ssa-structalias.c (associate_varinfo_to_alias): New function.
>> ? ? ? ?(ipa_pta_execute): Use it.
>>
>> ? ? ? ?* lto.c (add_cgraph_node_to_partition_1): Break out from ...
>> ? ? ? ?(add_cgraph_node_to_partition) ... here; walk aliases.
>> ? ? ? ?(lto_1_to_1_map): Remove same body alias code.
>> ? ? ? ?(promote_fn): Likewise.
>> ? ? ? ?(lto_promote_cross_file_statics): Update comment.
>>
>> ? ? ? ?* decl2.c (cp_write_global_declarations): Process aliases; look trhough
>> ? ? ? ?same body aliases.
>
> This caused:
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49373
>

This also pretty much destroyed C++ for ia32:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49378
http://gcc.gnu.org/ml/gcc-regression/2011-06/msg00159.html


-- 
H.J.