[PATCH] PR 84615 Regressions due to type mismatch with character functions

[PATCH] PR 84615 Regressions due to type mismatch with character functions

[Janne Blomqvist]

Since the kind of the hidden character length variable is not part of
the character variable definition, we must ensure that character
lengths are always of the same kind in interfaces, regardless of how
they were declared in the source. This patch ensures this when calling
a procedure.

Regtested on x86_64-pc-linux-gnu and i686-pc-linux-gnu, Ok for trunk?

gcc/fortran/ChangeLog:

2018-03-21  Janne Blomqvist  <jb@gcc.gnu.org>

	PR fortra/84615
	* trans-expr.c (gfc_conv_procedure_call): Convert charlen to
	gfc_charlen_type_node when calling procedure.

gcc/testsuite/ChangeLog:

2018-03-21  Janne Blomqvist  <jb@gcc.gnu.org>

	PR fortran/84615
	* gfortran.dg/char_result_17.f90: New test.
---
 gcc/fortran/trans-expr.c                     |  8 ++++++--
 gcc/testsuite/gfortran.dg/char_result_17.f90 | 20 ++++++++++++++++++++
 2 files changed, 26 insertions(+), 2 deletions(-)
 create mode 100644 gcc/testsuite/gfortran.dg/char_result_17.f90

diff --git a/gcc/fortran/trans-expr.c b/gcc/fortran/trans-expr.c
index 54bda1d..8bf5504 100644
--- a/gcc/fortran/trans-expr.c
+++ b/gcc/fortran/trans-expr.c
@@ -5973,9 +5973,13 @@ gfc_conv_procedure_call (gfc_se * se, gfc_symbol * sym,
 	  gfc_add_block_to_block (&se->pre, &parmse.pre);
 	  gfc_add_block_to_block (&se->post, &parmse.post);
 	  tmp = parmse.expr;
+	  /* TODO: It would be better to have the charlens as
+	     gfc_charlen_type_node already when the interface is
+	     created instead of converting it here (see PR 84615).  */
 	  tmp = fold_build2_loc (input_location, MAX_EXPR,
-				 TREE_TYPE (tmp), tmp,
-				 build_zero_cst (TREE_TYPE (tmp)));
+				 gfc_charlen_type_node,
+				 fold_convert (gfc_charlen_type_node, tmp),
+				 build_zero_cst (gfc_charlen_type_node));
 	  cl.backend_decl = tmp;
 	}
 
diff --git a/gcc/testsuite/gfortran.dg/char_result_17.f90 b/gcc/testsuite/gfortran.dg/char_result_17.f90
new file mode 100644
index 0000000..05ab72d
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/char_result_17.f90
@@ -0,0 +1,20 @@
+! { dg-do run }
+! PR fortran/84615
+! Charlen should always be the ABI defined character length type
+! regardless of which kind it is declared as in the source.
+program TestStringTools
+  character(len=52)               :: txt
+  character(len=1), dimension(52) :: chararr = &
+       (/(char(i+64),char(i+96), i = 1,26)/)
+  txt = chararray2string(chararr)
+  if (txt .ne. "AaBbCcDdEeFfGgHhIiJjKkLlMmNnOoPpQqRrSsTtUuVvWwXxYyZz") &
+       STOP 1
+contains
+  function chararray2string(chararray) result(text)
+    character(len=1), dimension(:) :: chararray    ! input
+    character(len=int(size(chararray, 1), kind=8)) :: text      ! output
+    do i = 1,size(chararray,1)
+       text(i:i) = chararray (i)
+    end do
+  end function chararray2string
+end program TestStringTools
-- 
2.7.4

Re: [PATCH] PR 84615 Regressions due to type mismatch with character functions

[Steve Kargl]

On Wed, Mar 21, 2018 at 10:39:06AM +0200, Janne Blomqvist wrote:
> Since the kind of the hidden character length variable is not part of
> the character variable definition, we must ensure that character
> lengths are always of the same kind in interfaces, regardless of how
> they were declared in the source. This patch ensures this when calling
> a procedure.
> 
> Regtested on x86_64-pc-linux-gnu and i686-pc-linux-gnu, Ok for trunk?
> 

OK.

-- 
Steve