[Bug c++/98364] New: C++20 module binary bloat

[Bug c++/98364] New: C++20 module binary bloat

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

            Bug ID: 98364
           Summary: C++20 module binary bloat
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: unlvsur at live dot com
  Target Milestone: ---

the module will generate _ZGIW5helloEv which is dead code. There is no way to
remove them.

naive binary without module:

#include<cstdio>

int main()
{
        puts("Hello World\n");
}

Disassembly of section .text:

0000000000401040 <main>:
  401040:       48 83 ec 08             sub    $0x8,%rsp
  401044:       bf 04 20 40 00          mov    $0x402004,%edi
  401049:       e8 e2 ff ff ff          callq  401030 <puts@plt>
  40104e:       31 c0                   xor    %eax,%eax
  401050:       48 83 c4 08             add    $0x8,%rsp
  401054:       c3                      retq   
  401055:       66 2e 0f 1f 84 00 00    nopw   %cs:0x0(%rax,%rax,1)
  40105c:       00 00 00 
  40105f:       90                      nop

0000000000401060 <set_fast_math>:
  401060:       f3 0f 1e fa             endbr64 
  401064:       0f ae 5c 24 fc          stmxcsr -0x4(%rsp)
  401069:       81 4c 24 fc 40 80 00    orl    $0x8040,-0x4(%rsp)
  401070:       00 
  401071:       0f ae 54 24 fc          ldmxcsr -0x4(%rsp)
  401076:       c3                      retq   
  401077:       66 0f 1f 84 00 00 00    nopw   0x0(%rax,%rax,1)
  40107e:       00 00 


module binary:

0000000000401050 <main>:
  401050:       48 83 ec 08             sub    $0x8,%rsp
  401054:       bf 04 20 40 00          mov    $0x402004,%edi
  401059:       e8 d2 ff ff ff          callq  401030 <puts@plt>
  40105e:       31 c0                   xor    %eax,%eax
  401060:       48 83 c4 08             add    $0x8,%rsp
  401064:       c3                      retq   
  401065:       66 66 2e 0f 1f 84 00    data16 nopw %cs:0x0(%rax,%rax,1)
  40106c:       00 00 00 00 

0000000000401070 <_GLOBAL__sub_I_main>:
  401070:       e9 0b 00 00 00          jmpq   401080 <_ZGIW5helloEv>
  401075:       66 2e 0f 1f 84 00 00    nopw   %cs:0x0(%rax,%rax,1)
  40107c:       00 00 00 
  40107f:       90                      nop

0000000000401080 <_ZGIW5helloEv>:
  401080:       80 3d b2 2f 00 00 00    cmpb   $0x0,0x2fb2(%rip)        #
404039 <_ZZ13_ZGIW5helloEvE9__in_chrg>
  401087:       75 07                   jne    401090 <_ZGIW5helloEv+0x10>
  401089:       c6 05 a9 2f 00 00 01    movb   $0x1,0x2fa9(%rip)        #
404039 <_ZZ13_ZGIW5helloEvE9__in_chrg>
  401090:       c3                      retq   
  401091:       66 2e 0f 1f 84 00 00    nopw   %cs:0x0(%rax,%rax,1)
  401098:       00 00 00 
  40109b:       0f 1f 44 00 00          nopl   0x0(%rax,%rax,1)

00000000004010a0 <set_fast_math>:
  4010a0:       f3 0f 1e fa             endbr64 
  4010a4:       0f ae 5c 24 fc          stmxcsr -0x4(%rsp)
  4010a9:       81 4c 24 fc 40 80 00    orl    $0x8040,-0x4(%rsp)
  4010b0:       00 
  4010b1:       0f ae 54 24 fc          ldmxcsr -0x4(%rsp)
  4010b6:       c3                      retq   
  4010b7:       66 0f 1f 84 00 00 00    nopw   0x0(%rax,%rax,1)
  4010be:       00 00

[Bug c++/98364] C++20 module global constructor emission

[nathan at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

Nathan Sidwell <nathan at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
           Assignee|unassigned at gcc dot gnu.org      |nathan at gcc dot gnu.org
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2020-12-18
            Summary|C++20 module binary bloat   |C++20 module global
                   |                            |constructor emission

--- Comment #1 from Nathan Sidwell <nathan at gcc dot gnu.org> ---
missed optimization/bad predicate

[Bug c++/98364] C++20 module global constructor emission

[nathan at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

Nathan Sidwell <nathan at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |WAITING

--- Comment #2 from Nathan Sidwell <nathan at gcc dot gnu.org> ---
Is this windows-specific? please privode preprocessed source and compilation
command.

[Bug c++/98364] C++20 module global constructor emission

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

--- Comment #3 from cqwrteur <unlvsur at live dot com> ---
Created attachment 49824
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49824&action=edit
just compile them and link together to see assembly.

[Bug c++/98364] C++20 module global constructor emission

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

--- Comment #4 from cqwrteur <unlvsur at live dot com> ---
(In reply to Nathan Sidwell from comment #2)
> Is this windows-specific? please privode preprocessed source and compilation
> command.

no

[Bug c++/98364] C++20 module global constructor emission

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

--- Comment #5 from cqwrteur <unlvsur at live dot com> ---
(In reply to cqwrteur from comment #3)
> Created attachment 49824 [details]
> just compile them and link together to see assembly.

compilation command

g++ -o main main.cc hello.cc -Ofast -std=c++20 -flto -fmodules-ts

[Bug c++/98364] C++20 module global constructor emission

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

--- Comment #6 from cqwrteur <unlvsur at live dot com> ---
(In reply to Nathan Sidwell from comment #2)
> Is this windows-specific? please privode preprocessed source and compilation
> command.

You need to just compile the code and then use objdump to view generated
assembly.

[Bug c++/98364] C++20 module global constructor emission

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

--- Comment #7 from cqwrteur <unlvsur at live dot com> ---
Created attachment 49825
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49825&action=edit
header only naive version

g++ -o naive naive.cc -Ofast -std=c++20 -s -flto

compile both of them and compare assembly.

[Bug c++/98364] [modules] unnneded global constructors are emitted for a module

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
            Summary|[modules] global            |[modules] unnneded global
                   |constructor emission        |constructors are emitted
                   |                            |for a module
           Severity|normal                      |enhancement

--- Comment #8 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed:
;; Function (static initializers for hello.cc) (_ZGIW5helloEv)
;; enabled by -tree-original


{
  static bool __in_chrg;

    static bool __in_chrg;
  if (__in_chrg)
    {
      return;
    }
  <<cleanup_point <<< Unknown tree: expr_stmt
  (void) (__in_chrg = 1) >>>>>;
}


In main-hello.cc.005t.original.

The code comes from start_objects in cp/decl2.c:


  if (module_init > 0)
    {
      // 'static bool __in_chrg = false;
      // if (__inchrg) return;
      // __inchrg = true
      tree var = build_lang_decl (VAR_DECL, in_charge_identifier,
                                  boolean_type_node);
      DECL_CONTEXT (var) = fndecl;
      DECL_ARTIFICIAL (var) = true;
      TREE_STATIC (var) = true;
      pushdecl (var);
      cp_finish_decl (var, NULL_TREE, false, NULL_TREE, 0);

      tree if_stmt = begin_if_stmt ();
      finish_if_stmt_cond (var, if_stmt);
      finish_return_stmt (NULL_TREE);
      finish_then_clause (if_stmt);
      finish_if_stmt (if_stmt);

      tree assign = build2 (MODIFY_EXPR, boolean_type_node,
                            var, boolean_true_node);
      TREE_SIDE_EFFECTS (assign) = true;
      finish_expr_stmt (assign);
    }
  if (module_init)
    module_add_import_initializers ();

Maybe if there are no initializers, maybe we should not create the global init.

[Bug c++/98364] [modules] unnneded global constructors are emitted for a module

[ppalka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98364

Patrick Palka <ppalka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ppalka at gcc dot gnu.org
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |13.0
         Resolution|---                         |FIXED

--- Comment #9 from Patrick Palka <ppalka at gcc dot gnu.org> ---
This seems fixed since r13-1090-ge8609768fbbc1e.