public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: "H.J. Lu" <hjl.tools@gmail.com>
To: Jan Hubicka <hubicka@ucw.cz>
Cc: gcc-patches@gcc.gnu.org
Subject: Re: Cgraph alias reorg 15/14 (New infrastructure for same body aliases)
Date: Sat, 11 Jun 2011 17:31:00 -0000	[thread overview]
Message-ID: <BANLkTim23fOqkBS+QKfmmpExBrAXfTNdAw@mail.gmail.com> (raw)
In-Reply-To: <20110611125552.GA25068@kam.mff.cuni.cz>

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


-- 
H.J.

  reply	other threads:[~2011-06-11 16:15 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-06-11 14:22 Jan Hubicka
2011-06-11 17:31 ` H.J. Lu [this message]
2011-06-12  9:02   ` H.J. Lu
2011-06-12 15:08     ` Jan Hubicka
2011-06-12 15:45       ` H.J. Lu
2011-06-12 15:55         ` Jan Hubicka
2011-06-12 16:33         ` H.J. Lu
2011-06-12 16:29       ` H.J. Lu
2011-06-12 17:06         ` Jan Hubicka
2011-06-12 17:46           ` H.J. Lu
2011-06-12 17:13         ` Jan Hubicka
2011-06-12 18:07           ` H.J. Lu
2011-06-12 22:11             ` Jan Hubicka
2011-06-19 12:52 ` Hans-Peter Nilsson
2011-06-19 12:55   ` Jan Hubicka
2011-06-19 14:04     ` Jan Hubicka
2011-06-20 11:23       ` Richard Guenther
2011-06-20 14:47         ` Hans-Peter Nilsson
2011-06-22 21:54       ` Hans-Peter Nilsson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=BANLkTim23fOqkBS+QKfmmpExBrAXfTNdAw@mail.gmail.com \
    --to=hjl.tools@gmail.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=hubicka@ucw.cz \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).