[Bug bootstrap/50689] New: AIX bootstrap failure from cgraphunit.c aliases change

[Bug bootstrap/50689] New: AIX bootstrap failure from cgraphunit.c aliases change

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

             Bug #: 50689
           Summary: AIX bootstrap failure from cgraphunit.c aliases change
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dje@gcc.gnu.org


Created attachment 25454
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25454
Pre-processed file that is miscompiled

The patch to cgraphunit.c forcing aliases to be output when cgraph decides so
causes an AIX bootstrap failure.  The patch shifts the output of the aliases in
the assembly file to earlier in the file, before the definition.  AIX assembler
is a single pass assembler and does not allow forward references.

[Bug bootstrap/50689] AIX bootstrap failure from cgraphunit.c aliases change

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

--- Comment #2 from David Edelsohn <dje at gcc dot gnu.org> 2011-10-10 19:08:13 UTC ---
Created attachment 25456
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25456
Assembler output of miscompiled file

[Bug bootstrap/50689] AIX bootstrap failure from cgraphunit.c aliases change

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

--- Comment #1 from David Edelsohn <dje at gcc dot gnu.org> 2011-10-10 19:07:43 UTC ---
Created attachment 25455
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25455
Assembler output of correctly compiled files

[Bug bootstrap/50689] AIX bootstrap failure from cgraphunit.c aliases change

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011-10-10
     Ever Confirmed|0                           |1

--- Comment #3 from David Edelsohn <dje at gcc dot gnu.org> 2011-10-10 19:13:52 UTC ---
In the file that works correctly, _ZNSt6localeD1Ev is aliased to
_ZNSt6localeD2Ev near the end of the file, after _ZNSt6localeD2Ev is defined. 
In the incorrect file, it is aliased early, just prior to the definition:

        .csect ._ZNSt6locale5_ImplD2Ev[PR],2
        .globl _ZNSt6localeD1Ev
        .globl  ._ZNSt6localeD1Ev
        .set    ._ZNSt6localeD1Ev,._ZNSt6localeD2Ev
        .set _ZNSt6localeD1Ev,_ZNSt6localeD2Ev
        .csect ._ZNSt6localeD2Ev[PR],2
        .align 2
        .globl _ZNSt6localeD2Ev
        .globl ._ZNSt6localeD2Ev
        .csect _ZNSt6localeD2Ev[DS]
_ZNSt6localeD2Ev:
        .long ._ZNSt6localeD2Ev, TOC[tc0], 0
        .csect ._ZNSt6localeD2Ev[PR],2
._ZNSt6localeD2Ev:
....

[Bug bootstrap/50689] AIX bootstrap failure from cgraphunit.c aliases change

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

--- Comment #4 from Jan Hubicka <hubicka at gcc dot gnu.org> 2011-10-14 12:50:54 UTC ---
Thanks. I am testing the following patch. Could you, please, verify that it
soves the problem on AIX?  I am quite convinced the aliases should be output
correctly with this change.
Index: cgraphunit.c
===================================================================
--- cgraphunit.c        (revision 179919)
+++ cgraphunit.c        (working copy)
@@ -1798,7 +1798,6 @@ cgraph_expand_function (struct cgraph_no

   announce_function (decl);
   node->process = 0;
-  assemble_thunks_and_aliases (node);
   gcc_assert (node->lowered);

   /* Generate RTL for the body of DECL.  */
@@ -1808,6 +1807,14 @@ cgraph_expand_function (struct cgraph_no
   gcc_assert (TREE_ASM_WRITTEN (decl));
   current_function_decl = NULL;
   gcc_assert (!cgraph_preserve_function_body_p (node));
+
+  /* It would make a lot more sense to output thunks before function body to
get more
+     forward and lest backwarding jumps.  This is however would need solving
problem
+     with comdats. See PR48668.  Also aliases must come after function itself
to
+     make one pass assemblers, like one on AIX happy.  See PR 50689.
+     FIXME: Perhaps thunks should be move before function IFF they are not in
comdat
+     groups.  */
+  assemble_thunks_and_aliases (node);
   cgraph_release_function_body (node);
   /* Eliminate all call edges.  This is important so the GIMPLE_CALL no longer
      points to the dead function body.  */

[Bug bootstrap/50689] AIX bootstrap failure from cgraphunit.c aliases change

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

--- Comment #5 from David Edelsohn <dje at gcc dot gnu.org> 2011-10-14 18:47:15 UTC ---
Bootstrap is not complete, but it is past the location of the failure with your
patch applied.

Thanks, David

[Bug bootstrap/50689] AIX bootstrap failure from cgraphunit.c aliases change

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

--- Comment #6 from Jan Hubicka <hubicka at gcc dot gnu.org> 2011-10-16 09:02:38 UTC ---
Author: hubicka
Date: Sun Oct 16 09:02:33 2011
New Revision: 180053

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=180053
Log:

    PR target/48668
    PR target/50689
    * cgraphunit.c (cgraph_expand_function): Expand thunks and alises
    after function body.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cgraphunit.c

[Bug bootstrap/50689] AIX bootstrap failure from cgraphunit.c aliases change

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50689

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #7 from David Edelsohn <dje at gcc dot gnu.org> ---
Fixed