[patch, fortran] Bug 81296 - derived type I/o problem

[patch, fortran] Bug 81296 - derived type I/o problem

[Jerry DeLisle]

[-- Attachment #1: Type: text/plain, Size: 499 bytes --]

Hi all,

The attached patch adds a check for the format label containing a "DT" format 
descriptor and enables the generation of the correct code.  The patch modifies 
an existing test case as a future check on this.

Regression tested on x86_64-linux.

OK for trunk and backport to 7?

Regards,

Jerry

2017-08-21  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

	PR fortran/81296
	* trans-io.c (get_dtio_proc): Add check for format label and set
	formatted flag accordingly. Reorganize the code a little.

[-- Attachment #2: pr81296.diff --]
[-- Type: text/x-patch, Size: 1718 bytes --]

diff --git a/gcc/fortran/trans-io.c b/gcc/fortran/trans-io.c
index c3c56f29..aa974eb3 100644
--- a/gcc/fortran/trans-io.c
+++ b/gcc/fortran/trans-io.c
@@ -2214,18 +2214,24 @@ get_dtio_proc (gfc_typespec * ts, gfc_code * code, gfc_symbol **dtio_sub)
   bool formatted = false;
   gfc_dt *dt = code->ext.dt;
 
-  if (dt && dt->format_expr)
+  if (dt)
     {
-      char *fmt;
-      fmt = gfc_widechar_to_char (dt->format_expr->value.character.string,
-				  -1);
-      if (strtok (fmt, "DT") != NULL)
+      char *fmt = NULL;
+
+      if (dt->format_label == &format_asterisk)
+	{
+	  /* List directed io must call the formatted DTIO procedure.  */
+	  formatted = true;
+	}
+      else if (dt->format_expr)
+	fmt = gfc_widechar_to_char (dt->format_expr->value.character.string,
+				      -1);
+      else if (dt->format_label)
+	fmt = gfc_widechar_to_char (dt->format_label->format->value.character.string,
+				      -1);
+      if (fmt && strtok (fmt, "DT") != NULL)
 	formatted = true;
-    }
-  else if (dt && dt->format_label == &format_asterisk)
-    {
-      /* List directed io must call the formatted DTIO procedure.  */
-      formatted = true;
+
     }
 
   if (ts->type == BT_CLASS)
diff --git a/gcc/testsuite/gfortran.dg/dtio_12.f90 b/gcc/testsuite/gfortran.dg/dtio_12.f90
index 213f7ebb..cf1bfe38 100644
--- a/gcc/testsuite/gfortran.dg/dtio_12.f90
+++ b/gcc/testsuite/gfortran.dg/dtio_12.f90
@@ -70,5 +70,11 @@ end module
   rewind (10)
   read (10, *) msg
   if (trim (msg) .ne. "77") call abort
+  rewind (10)
+  write (10,40) child (77)                         ! Modified using format label
+40 format(DT)
+  rewind (10)
+  read (10, *) msg
+  if (trim (msg) .ne. "77") call abort
   close(10)
 end

Re: [patch, fortran] Bug 81296 - derived type I/o problem

[Paul Richard Thomas]

Hi Jerry,

That looks good to me - OK for trunk and for backporting.

Thanks for the patch.

Paul

PS Did you have time to think about that rather more difficult bug,
involving a mix of DT descriptors and intrinsic descriptors for the
declared type?

On 21 August 2017 at 03:23, Jerry DeLisle <jvdelisle@charter.net> wrote:
> Hi all,
>
> The attached patch adds a check for the format label containing a "DT"
> format descriptor and enables the generation of the correct code.  The patch
> modifies an existing test case as a future check on this.
>
> Regression tested on x86_64-linux.
>
> OK for trunk and backport to 7?
>
> Regards,
>
> Jerry
>
> 2017-08-21  Jerry DeLisle  <jvdelisle@gcc.gnu.org>
>
>         PR fortran/81296
>         * trans-io.c (get_dtio_proc): Add check for format label and set
>         formatted flag accordingly. Reorganize the code a little.



-- 
"If you can't explain it simply, you don't understand it well enough"
- Albert Einstein