diff mbox

[Fortran] Fix PR 71902

Message ID a64bf1ca-af8d-1912-1e2a-8ef2a0c5ecec@netcologne.de
State New
Headers show

Commit Message

Thomas Koenig Aug. 14, 2016, 1:05 p.m. UTC 
Hello world,

this patch fixes the performance regression introduced with
the fix for 71783.  It also corrects an error in the logic
in realloc_string_callback by now checking if the experssion
is deferred, instead of checking for the symbol having the
allocatable attribute (which led to false positives).

It also checks for the presence of a substring on the RHS
expression, because the problem cannot happen if there is
no RHS substring.

Regression-tested on trunk.

OK for trunk?  Also OK for a backport?

Regards

	Thomas

2016-08-14  Thomas Koenig  <tkoenig@gcc.gnu.org>

         PR fortran/71902
         * frontend-passes.c (realloc_string_callback):  Check for deferred
         on the expression instead for allocatable on the symbol.  Name 
temporary
         variable "realloc_string".

2016-08-14  Thomas Koenig  <tkoenig@gcc.gnu.org>

         PR fortran/71902
         * gfortran.dg/dependency_47.f90:  New test.

Comments

Mikael Morin Aug. 16, 2016, 7:07 p.m. UTC | #1 
Hello,

Le 14/08/2016 à 15:05, Thomas Koenig a écrit :
> Index: frontend-passes.c
> ===================================================================
> --- frontend-passes.c	(Revision 239218)
> +++ frontend-passes.c	(Arbeitskopie)
> @@ -177,6 +178,13 @@ realloc_string_callback (gfc_code **c, int *walk_s
>    if (expr2->expr_type != EXPR_VARIABLE)
>      return 0;
>
> +  /* Only substring expressions can be affected; substrings are always the
> +   last reference.  */
> +
> +  for (ref = expr2->ref; ref; ref = ref->next)
> +    if (ref->type != REF_SUBSTRING)
> +      return 0;
> +
>    if (!gfc_check_dependency (expr1, expr2, true))
>      return 0;
>
This doesn't look correct, what about substrings following component or 
array references?

Mikael

PS: What about the original fix, wasn't it a dependency 
problem/inaccuracy after all?
diff mbox

Patch

Index: frontend-passes.c
===================================================================
--- frontend-passes.c	(Revision 239218)
+++ frontend-passes.c	(Arbeitskopie)
@@ -164,6 +164,7 @@  realloc_string_callback (gfc_code **c, int *walk_s
   gfc_expr *expr1, *expr2;
   gfc_code *co = *c;
   gfc_expr *n;
+  gfc_ref *ref;
 
   if (co->op != EXEC_ASSIGN)
     return 0;
@@ -170,7 +171,7 @@  realloc_string_callback (gfc_code **c, int *walk_s
 
   expr1 = co->expr1;
   if (expr1->ts.type != BT_CHARACTER || expr1->rank != 0
-      || !expr1->symtree->n.sym->attr.allocatable)
+      || !expr1->ts.deferred)
     return 0;
 
   expr2 = gfc_discard_nops (co->expr2);
@@ -177,6 +178,13 @@  realloc_string_callback (gfc_code **c, int *walk_s
   if (expr2->expr_type != EXPR_VARIABLE)
     return 0;
 
+  /* Only substring expressions can be affected; substrings are always the
+   last reference.  */
+
+  for (ref = expr2->ref; ref; ref = ref->next)
+    if (ref->type != REF_SUBSTRING)
+      return 0;
+
   if (!gfc_check_dependency (expr1, expr2, true))
     return 0;
 
@@ -190,7 +198,7 @@  realloc_string_callback (gfc_code **c, int *walk_s
   current_code = c;
   inserted_block = NULL;
   changed_statement = NULL;
-  n = create_var (expr2, "trim");
+  n = create_var (expr2, "realloc_string");
   co->expr2 = n;
   return 0;
 }