[Patch, fortran] PR64952 - Missing temporary in assignment from elemental function

[Mikael Morin]

Hello Paul,

comments below

Le 08/02/2015 16:24, Paul Richard Thomas a écrit :
> 
> Index: gcc/fortran/gfortran.h
> ===================================================================
> *** gcc/fortran/gfortran.h	(revision 220482)
> --- gcc/fortran/gfortran.h	(working copy)
> *************** typedef struct
> *** 789,794 ****
> --- 789,798 ----
>        cannot alias.  Note that this is zero for PURE procedures.  */
>     unsigned implicit_pure:1;
>   
> +   /* This set for an elemental function that contains expressions for
> +      arrays coming from outside its namespace.  */
> +   unsigned potentially_aliased:1;
> + 
aliased is more something about pointers, so how about naming it
something like array_outer_dependency?  Anyway, that's minor.

I wonder whether we should negate the meaning, that is set the flag if
there is no external dependency.
If we can get the conditions to set it exhaustively right, both are
equivalent.  Otherwise... maybe not.

>     /* This is set if the subroutine doesn't return.  Currently, this
>        is only possible for intrinsic subroutines.  */
>     unsigned noreturn:1;
> Index: gcc/fortran/trans.h
> ===================================================================
> *** gcc/fortran/trans.h	(revision 220481)
> --- gcc/fortran/trans.h	(working copy)
> *************** typedef struct gfc_ss_info
> *** 226,231 ****
> --- 226,235 ----
>     /* Suppresses precalculation of scalars in WHERE assignments.  */
>     unsigned where:1;
>   
> +   /* Signals that an array argument of an elemental function might be aliased,
> +      thereby generating a temporary in assignments.  */
> +   unsigned potentially_aliased:1;
> + 
>     /* Tells whether the SS is for an actual argument which can be a NULL
>        reference.  In other words, the associated dummy argument is OPTIONAL.
>        Used to handle elemental procedures.  */
> Index: gcc/fortran/resolve.c
> ===================================================================
> *** gcc/fortran/resolve.c	(revision 220481)
> --- gcc/fortran/resolve.c	(working copy)
> *************** resolve_variable (gfc_expr *e)
> *** 5054,5059 ****
> --- 5054,5067 ----
>   		    && gfc_current_ns->parent->parent == sym->ns)))
>       sym->attr.host_assoc = 1;
>   
> +   if (sym->attr.dimension
> +       && (sym->ns != gfc_current_ns
> + 	  || sym->attr.use_assoc
> + 	  || sym->attr.in_common)
> +       && gfc_elemental (NULL)
> +       && gfc_current_ns->proc_name->attr.function)
> +     gfc_current_ns->proc_name->attr.potentially_aliased = 1;
I would expect the flag to also be copied between procedures in some
cases; namely if A calls B, and B has the flag, then A has the flag.
There is also the case of external procedures (for which the flag is not
known -> assume the worst)

> + 
>   resolve_procedure:
>     if (t && !resolve_procedure_expression (e))
>       t = false;
> Index: gcc/fortran/trans-array.c
> ===================================================================
> *** gcc/fortran/trans-array.c	(revision 220482)
> --- gcc/fortran/trans-array.c	(working copy)
> *************** gfc_conv_resolve_dependencies (gfc_loopi
> *** 4391,4396 ****
> --- 4391,4402 ----
>       {
>         ss_expr = ss->info->expr;
>   
> +       if (ss->info->potentially_aliased)
> + 	{
> + 	  nDepend = 1;
> + 	  break;
> + 	}
> + 
>         if (ss->info->type != GFC_SS_SECTION)
>   	{
>   	  if (flag_realloc_lhs
> *************** gfc_walk_function_expr (gfc_ss * ss, gfc
> *** 9096,9104 ****
>     /* Walk the parameters of an elemental function.  For now we always pass
>        by reference.  */
>     if (sym->attr.elemental || (comp && comp->attr.elemental))
> !     return gfc_walk_elemental_function_args (ss, expr->value.function.actual,
>   					     gfc_get_proc_ifc_for_expr (expr),
>   					     GFC_SS_REFERENCE);
>   
>     /* Scalar functions are OK as these are evaluated outside the scalarization
>        loop.  Pass back and let the caller deal with it.  */
> --- 9102,9114 ----
>     /* Walk the parameters of an elemental function.  For now we always pass
>        by reference.  */
>     if (sym->attr.elemental || (comp && comp->attr.elemental))
> !     {
> !       ss = gfc_walk_elemental_function_args (ss, expr->value.function.actual,
>   					     gfc_get_proc_ifc_for_expr (expr),
>   					     GFC_SS_REFERENCE);
> +       if (sym->attr.potentially_aliased)
> + 	ss->info->potentially_aliased = 1;
> +     }

This is somewhat hackish, potentially_aliased is a global thing, not
specific to SS, and this may end up marking gfc_ss_terminator as
potentiallly_aliased for example, but I don't see any other obvious way
to do it, so it's OK I guess.

Anyway, the comp && comp->attr.elemental part of the if should be
handled too (always set the flag in that case?).  I actually wonder why
it works without.

I attach a few variants of the testcase, which don't work yet.

Mikael

-------------- next part --------------
A non-text attachment was scrubbed...
Name: test.f90
Type: text/x-fortran
Size: 497 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20150208/a95025a4/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test2.f90
Type: text/x-fortran
Size: 998 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20150208/a95025a4/attachment-0001.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test3.f90
Type: text/x-fortran
Size: 599 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20150208/a95025a4/attachment-0002.bin>