public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [patch, fortran] Fix PR 80988
@ 2017-06-09 20:47 Thomas Koenig
  2017-06-09 23:45 ` Jerry DeLisle
  0 siblings, 1 reply; 2+ messages in thread
From: Thomas Koenig @ 2017-06-09 20:47 UTC (permalink / raw)
  To: fortran, gcc-patches

[-- Attachment #1: Type: text/plain, Size: 558 bytes --]

Hello world,

the attached patch fixes the PR by not doing a replacement of
(a(i,i),i=1,3) by an array expression (which does not exist).

Regression-tested.  OK for trunk?

Regards

	Thomas

2017-06-09  Thomas Koenig  <tkoenig@gcc.gnu.org>

         PR fortran/80988
         * frontend-passes.c (traverse_io_block):  Also
         check for variables occurring as indices multiple
         time in a single implied DO loop.

2017-06-09  Thomas Koenig  <tkoenig@gcc.gnu.org>

         PR fortran/80988
         * gfortran.dg/implied_do_io_3.f90:  New test.

[-- Attachment #2: p-swim.diff --]
[-- Type: text/x-patch, Size: 678 bytes --]

Index: frontend-passes.c
===================================================================
--- frontend-passes.c	(Revision 249039)
+++ frontend-passes.c	(Arbeitskopie)
@@ -1164,7 +1164,14 @@ traverse_io_block (gfc_code *code, bool *has_reach
 	  /*  Check for (a(k), i=1,4) or ((a(j, i), i=1,4), j=1,4).  */
 	  if (!stack_top || !stack_top->iter
 	      || stack_top->iter->var->symtree != start->symtree)
-	    iters[i] = NULL;
+	    {
+	      /* Check for (a(i,i), i=1,3).  */
+	      for (j=0; j<i; j++)
+		if (iters[j] && iters[j]->var->symtree == start->symtree)
+		  return false;
+
+	      iters[i] = NULL;
+	    }
 	  else
 	    {
 	      iters[i] = stack_top->iter;

[-- Attachment #3: implied_do_io_3.f90 --]
[-- Type: text/x-fortran, Size: 794 bytes --]

! { dg-do run }
! { dg-options "-ffrontend-optimize" }
! PR 80988 - implied do loops with diagonal elements
! were not written correctly
program main
  implicit none
  integer :: i,j,k
  integer, dimension(3,3) :: a
  integer, dimension(3,3,3) :: b
  character(len=40) :: line
  a = reshape([(((i*10+j),i=1,3),j=1,3)], shape(a))
  i = 2147483548
  write (unit=line,fmt='(10I3)') (a(i,i),i=1,3)
  if (line /= ' 11 22 33') call abort
  write (unit=line,fmt='(10I3)') (a(i+1,i+1),i=1,2)
  if (line /= ' 22 33') call abort
  do k=1,3
     do j=1,3
        do i=1,3
           b(i,j,k) = i*100 + j*10 + k
        end do
     end do
  end do
  i = -2147483548
  write (unit=line,fmt='(10I4)') ((b(i,j,i),i=1,3),j=1,3)
  if (line /= ' 111 212 313 121 222 323 131 232 333') call abort
end program main

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [patch, fortran] Fix PR 80988
  2017-06-09 20:47 [patch, fortran] Fix PR 80988 Thomas Koenig
@ 2017-06-09 23:45 ` Jerry DeLisle
  0 siblings, 0 replies; 2+ messages in thread
From: Jerry DeLisle @ 2017-06-09 23:45 UTC (permalink / raw)
  To: Thomas Koenig, fortran, gcc-patches

On 06/09/2017 01:47 PM, Thomas Koenig wrote:
> Hello world,
> 
> the attached patch fixes the PR by not doing a replacement of
> (a(i,i),i=1,3) by an array expression (which does not exist).
> 
> Regression-tested.  OK for trunk?
>

OK, thanks

> Regards
> 
>      Thomas
> 
> 2017-06-09  Thomas Koenig  <tkoenig@gcc.gnu.org>
> 
>          PR fortran/80988
>          * frontend-passes.c (traverse_io_block):  Also
>          check for variables occurring as indices multiple
>          time in a single implied DO loop.
> 
> 2017-06-09  Thomas Koenig  <tkoenig@gcc.gnu.org>
> 
>          PR fortran/80988
>          * gfortran.dg/implied_do_io_3.f90:  New test.

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2017-06-09 23:45 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-06-09 20:47 [patch, fortran] Fix PR 80988 Thomas Koenig
2017-06-09 23:45 ` Jerry DeLisle

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).