[patch, fortran] Fix PR 50627

[patch, fortran] Fix PR 50627

[Thomas Koenig]

[-- Attachment #1: Type: text/plain, Size: 671 bytes --]

Hello world,

the attached patch fixes the regression regarding freeing
namespaces twice.

The test cases where also run unter valgrind without
producing errors.

Regression-tested.  OK for trunk?

	Thomas

2013-01-27  Thomas Koenig  <tkoenig@gcc.gnu.org>

         PR fortran/56054
         * decl.c (gfc_match_end):  Remove half-ready namespace
         from parent if the end of a block is missing.
         * parse.c (parse_module):  Do not put namespace into
         gsymbol on error.

2013-01-27  Thomas Koenig  <tkoenig@gcc.gnu.org>

         PR fortran/56054
         * gfortran.dg/block_12.f90:  New test.
         * gfortran.dg/module_error_1.f90:  New test.


[-- Attachment #2: p3.diff --]
[-- Type: text/x-patch, Size: 1965 bytes --]

Index: decl.c
===================================================================
--- decl.c	(Revision 195319)
+++ decl.c	(Arbeitskopie)
@@ -5979,6 +5979,8 @@ gfc_match_end (gfc_statement *st)
   const char *target;
   int eos_ok;
   match m;
+  gfc_namespace *parent_ns, *ns, *prev_ns;
+  gfc_namespace **nsp;
 
   old_loc = gfc_current_locus;
   if (gfc_match ("end") != MATCH_YES)
@@ -6204,6 +6206,35 @@ syntax:
 
 cleanup:
   gfc_current_locus = old_loc;
+
+  /* If we are missing an END BLOCK, we created a half-ready namespace.
+     Remove it from the parent namespace's sibling list.  */
+
+  if (state == COMP_BLOCK)
+    {
+      parent_ns = gfc_current_ns->parent;
+
+      nsp = &(gfc_state_stack->previous->tail->ext.block.ns);
+
+      prev_ns = NULL;
+      ns = *nsp;
+      while (ns)
+	{
+	  if (ns == gfc_current_ns)
+	    {
+	      if (prev_ns == NULL)
+		*nsp = NULL;
+	      else
+		prev_ns->sibling = ns->sibling;
+	    }
+	  prev_ns = ns;
+	  ns = ns->sibling;
+	}
+  
+      gfc_free_namespace (gfc_current_ns);
+      gfc_current_ns = parent_ns;
+    }
+
   return MATCH_ERROR;
 }
 
Index: parse.c
===================================================================
--- parse.c	(Revision 195319)
+++ parse.c	(Arbeitskopie)
@@ -4291,6 +4291,7 @@ parse_module (void)
 {
   gfc_statement st;
   gfc_gsymbol *s;
+  bool error;
 
   s = gfc_get_gsymbol (gfc_new_block->name);
   if (s->defined || (s->type != GSYM_UNKNOWN && s->type != GSYM_MODULE))
@@ -4304,6 +4305,7 @@ parse_module (void)
 
   st = parse_spec (ST_NONE);
 
+  error = false;
 loop:
   switch (st)
     {
@@ -4322,12 +4324,15 @@ loop:
       gfc_error ("Unexpected %s statement in MODULE at %C",
 		 gfc_ascii_statement (st));
 
+      error = true;
       reject_statement ();
       st = next_statement ();
       goto loop;
     }
 
-  s->ns = gfc_current_ns;
+  /* Make sure not to free the namespace twice on error.  */
+  if (!error)
+    s->ns = gfc_current_ns;
 }
 
 

[-- Attachment #3: module_error_1.f90 --]
[-- Type: text/x-fortran, Size: 119 bytes --]

! { dg-do compile }
module kernels
      select type (args) ! { dg-error "Unexpected SELECT TYPE" }
end module kernels

[-- Attachment #4: block_12.f90 --]
[-- Type: text/x-fortran, Size: 182 bytes --]

! { dg-do compile }
! PR 56054 - this used to free a namespace twice.
program main
  block
end program main ! { dg-error "END BLOCK" }
! { dg-prune-output "Unexpected end of file" }

Re: [patch, fortran] Fix PR 50627

[Thomas Koenig]

Of course, that should be PR 50627 in the ChangeLog.

Re: [patch, fortran] Fix PR 50627

[Paul Richard Thomas]

Dear Thomas,

I wrote this almost immediately after you posted the patch, got
disturbed and forgot about it - my apologies.  Note Dominique's reply
after you corrected the PR number.

OK for trunk.  The part in decl.c is exactly what I was thinking to do.

Thanks for the patch

Paul

Re: [patch, fortran] Fix PR 50627

[Thomas Koenig]

Hi Paul,

> OK for trunk.  The part in decl.c is exactly what I was thinking to do.
>
> Thanks for the patch

Thanks a lot for your review!

Committed to trunk as rev. 195684, 4.7 and 4.6 to follow.

	Thomas

Re: [patch, fortran] Fix PR 50627

[Dominique Dhumieres]

> Of course, that should be PR 50627 in the ChangeLog.

I think you need both PR 50627 and 56054.

The patch fixes these PRs, full testing in progress.

Dominique