[Bug plugins/56622] New: -fopenmp with -fplugin conflicts

[Bug plugins/56622] New: -fopenmp with -fplugin conflicts

[hayder.alkhalissi at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56622

             Bug #: 56622
           Summary: -fopenmp with -fplugin conflicts
    Classification: Unclassified
           Product: gcc
           Version: 4.5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: plugins
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hayder.alkhalissi@googlemail.com


Dear all,

I am trying to extending the compiler gcc-4.5.0 by using plug-in. When I tried
using -fpluging with -fopenmp, I got this error as bellow:

*** WARNING *** there are active plugins, do not report this as a bug unless
you can reproduce it without enabling any plugins.
Event                            | Plugins
PLUGIN_PRE_GENERICIZE            | libplugin-pointer-types
PLUGIN_FINISH                    | libplugin-pointer-types
PLUGIN_ATTRIBUTES                | libplugin-pointer-types
pointer-types-target.c: In function ‘main’:
pointer-types-target.c:46:2: internal compiler error: in
join_on_bb_function_call, at aop-pc-fun-call.c:246
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

BTY: I didn't have more experience in extending gcc.

I have already attached the *.i file as well.

[Bug plugins/56622] -fopenmp with -fplugin conflicts

[hayder.alkhalissi at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56622

--- Comment #1 from hayder.alkhalissi at googlemail dot com 2013-03-14 18:55:13 UTC ---
Created attachment 29670
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29670
preprocessed file

[Bug plugins/56622] -fopenmp with -fplugin conflicts

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56622

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> 2013-03-14 20:30:13 UTC ---
>join_on_bb_function_call, at aop-pc-fun-call.c:246

This is a bug in your plugin code and not in GCC.  Please report it to the
place where you got the plugin.

[Bug plugins/56622] -fopenmp with -fplugin conflicts

[hayder.alkhalissi at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56622

--- Comment #3 from hayder.alkhalissi at googlemail dot com 2013-03-15 08:35:29 UTC ---
(In reply to comment #2)
> >join_on_bb_function_call, at aop-pc-fun-call.c:246
> 
> This is a bug in your plugin code and not in GCC.  Please report it to the
> place where you got the plugin.

Thanks for fast response.
I found the code in aop_pc_fun_call.c:

gimple stmt = gsi_stmt (gsi)  /* gimple_stmt_iterator gsi; */

/* At this stage, there should be no GIMPLE statements with sub-statements. */
gcc-assert(!gimple_has_substatements (stmt));       <------line 246

As explained above, the error in using  gcc-assert().

Do you know why? and how could avoided?

Thanks

[Bug plugins/56622] -fopenmp with -fplugin conflicts

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56622

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> 2013-03-15 08:57:18 UTC ---
openMP is lowered later than your plugin runs and there are still
sub-statements
around.