diff mbox

[Fortran,pr78958,v1] Unallocated memory access after SOURCE-ALLOCATEing unlimited polymorphic object

Message ID 20170205134305.7793b284@vepi2
State New
Headers show

Commit Message

Andre Vehreschild Feb. 5, 2017, 12:43 p.m. UTC 
Hi Paul,

thanks for the review. Committed as r245192.

Regards,
	Andre

On Sat, 4 Feb 2017 17:07:13 +0000
Paul Richard Thomas <paul.richard.thomas@gmail.com> wrote:

> Hi Andre,
> 
> This looks to be OK for trunk.
> 
> Thanks for the patch.
> 
> Paul
> 
> On 4 February 2017 at 13:11, Andre Vehreschild <vehre@gmx.de> wrote:
> > Hi all,
> >
> > attached patch takes the _len-component of unlimited polymorphic objects
> > into account, when source-ALLOCATEing an unlimited polymorphic object. The
> > testcase for this gcc/testsuite/gfortran.dg/alloc_comp_class_5.f03 with
> > valgrind/sanitizer. I therefore added no additional testcase.
> >
> > Bootstraps and regtests ok on x86_64-linux/f25 and as reported in the PR on
> > 32bit-hpux. Ok for trunk (when?)?
> >
> > Regards,
> >         Andre
> > --
> > Andre Vehreschild * Email: vehre ad gmx dot de  
> 
> 
>
diff mbox

Patch

Index: gcc/fortran/ChangeLog
===================================================================
--- gcc/fortran/ChangeLog	(Revision 245191)
+++ gcc/fortran/ChangeLog	(Arbeitskopie)
@@ -1,5 +1,11 @@ 
 2017-02-05  Andre Vehreschild  <vehre@gcc.gnu.org>
 
+	PR fortran/78958
+	* trans-stmt.c (gfc_trans_allocate): Add the multiplying the _len
+	component of unlimited polymorphic objects when source-allocating.
+
+2017-02-05  Andre Vehreschild  <vehre@gcc.gnu.org>
+
 	PR fortran/79230
 	* trans-array.c (structure_alloc_comps): Ignore pointer components when
 	freeing structures.
Index: gcc/fortran/trans-stmt.c
===================================================================
--- gcc/fortran/trans-stmt.c	(Revision 245191)
+++ gcc/fortran/trans-stmt.c	(Arbeitskopie)
@@ -6009,14 +6009,21 @@ 
 	 needs to be provided, which is done most of the time by the
 	 pre-evaluation step.  */
       nelems = NULL_TREE;
-      if (expr3_len && code->expr3->ts.type == BT_CHARACTER)
-	/* When al is an array, then the element size for each element
-	   in the array is needed, which is the product of the len and
-	   esize for char arrays.  */
-	tmp = fold_build2_loc (input_location, MULT_EXPR,
-			       TREE_TYPE (expr3_esize), expr3_esize,
-			       fold_convert (TREE_TYPE (expr3_esize),
-					     expr3_len));
+      if (expr3_len && (code->expr3->ts.type == BT_CHARACTER
+			|| code->expr3->ts.type == BT_CLASS))
+	{
+	  /* When al is an array, then the element size for each element
+	     in the array is needed, which is the product of the len and
+	     esize for char arrays.  For unlimited polymorphics len can be
+	     zero, therefore take the maximum of len and one.  */
+	  tmp = fold_build2_loc (input_location, MAX_EXPR,
+				 TREE_TYPE (expr3_len),
+				 expr3_len, fold_convert (TREE_TYPE (expr3_len),
+							  integer_one_node));
+	  tmp = fold_build2_loc (input_location, MULT_EXPR,
+				 TREE_TYPE (expr3_esize), expr3_esize,
+				 fold_convert (TREE_TYPE (expr3_esize), tmp));
+	}
       else
 	tmp = expr3_esize;
       if (!gfc_array_allocate (&se, expr, stat, errmsg, errlen,