public inbox for fortran@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH] Fortran: fix simplification of intrinsics IBCLR and IBSET [PR106557]
@ 2022-08-20 19:11 Harald Anlauf
  2022-08-22  9:52 ` Mikael Morin
  0 siblings, 1 reply; 2+ messages in thread
From: Harald Anlauf @ 2022-08-20 19:11 UTC (permalink / raw)
  To: fortran, gcc-patches

[-- Attachment #1: Type: text/plain, Size: 912 bytes --]

Dear all,

the simplification of the TRANSFER intrinsic produces a
redundant representation of the result, one in expr->value
and another in expr->representation.string.  This is done
to ensure a safe "round-trip" for nested TRANSFER.

In a subsequent use of this result we either need to make
sure that both parts stay consistent, or drop the latter
part.

The simplifications of IBCLR and IBSET do a gfc_copy_expr
of their argument x and modify only the former part.
Depending on context, either value.integer or
representation.string (if non-NULL) are used later,
leading to surprising results.

A conservative approach to fix this PR is to simply
drop the unneeded representation.string in the
simplification of the intrinsics IBCLR and IBSET,
see attached patch.  (A quick glance did not turn up
other intrinsics affected the same way.)

Regtested on x86_64-pc-linux-gnu.  OK for mainline?

Thanks,
Harald


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Fortran-fix-simplification-of-intrinsics-IBCLR-and-I.patch --]
[-- Type: text/x-patch, Size: 2849 bytes --]

From a540a806cf02d739a408f129738252e73f03e60c Mon Sep 17 00:00:00 2001
From: Harald Anlauf <anlauf@gmx.de>
Date: Sat, 20 Aug 2022 20:36:28 +0200
Subject: [PATCH] Fortran: fix simplification of intrinsics IBCLR and IBSET
 [PR106557]

gcc/fortran/ChangeLog:

	PR fortran/106557
	* simplify.cc (gfc_simplify_ibclr): Ensure consistent results of
	the simplification by dropping a redundant memory representation
	of argument x.
	(gfc_simplify_ibset): Likewise.

gcc/testsuite/ChangeLog:

	PR fortran/106557
	* gfortran.dg/pr106557.f90: New test.
---
 gcc/fortran/simplify.cc                | 14 ++++++++++++++
 gcc/testsuite/gfortran.dg/pr106557.f90 | 19 +++++++++++++++++++
 2 files changed, 33 insertions(+)
 create mode 100644 gcc/testsuite/gfortran.dg/pr106557.f90

diff --git a/gcc/fortran/simplify.cc b/gcc/fortran/simplify.cc
index fb725994653..f992c31e5d7 100644
--- a/gcc/fortran/simplify.cc
+++ b/gcc/fortran/simplify.cc
@@ -3380,6 +3380,13 @@ gfc_simplify_ibclr (gfc_expr *x, gfc_expr *y)
   k = gfc_validate_kind (x->ts.type, x->ts.kind, false);

   result = gfc_copy_expr (x);
+  /* Drop any separate memory representation of x to avoid potential
+     inconsistencies in result.  */
+  if (result->representation.string)
+    {
+      free (result->representation.string);
+      result->representation.string = NULL;
+    }

   convert_mpz_to_unsigned (result->value.integer,
 			   gfc_integer_kinds[k].bit_size);
@@ -3471,6 +3478,13 @@ gfc_simplify_ibset (gfc_expr *x, gfc_expr *y)
   k = gfc_validate_kind (x->ts.type, x->ts.kind, false);

   result = gfc_copy_expr (x);
+  /* Drop any separate memory representation of x to avoid potential
+     inconsistencies in result.  */
+  if (result->representation.string)
+    {
+      free (result->representation.string);
+      result->representation.string = NULL;
+    }

   convert_mpz_to_unsigned (result->value.integer,
 			   gfc_integer_kinds[k].bit_size);
diff --git a/gcc/testsuite/gfortran.dg/pr106557.f90 b/gcc/testsuite/gfortran.dg/pr106557.f90
new file mode 100644
index 00000000000..d073f3e7186
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/pr106557.f90
@@ -0,0 +1,19 @@
+! { dg-do run }
+! { dg-additional-options "-fdump-tree-original" }
+! PR fortran/106557 - nesting intrinsics ibset and transfer gives wrong result
+
+program p
+  implicit none
+  character(1) :: s
+
+  write(s,'(i1)') ibset (transfer (0, 0), 0)
+  if (s /= '1') stop 1
+  write(s,'(i1)') ibclr (transfer (1, 0), 0)
+  if (s /= '0') stop 2
+
+  ! These shall be fully resolved at compile time:
+  if (transfer   (ibset (transfer (0, 0), 0), 0) /= 1) stop 3
+  if (transfer   (ibclr (transfer (1, 0), 0), 0) /= 0) stop 4
+end
+
+! { dg-final { scan-tree-dump-times "_gfortran_stop_numeric" 2 "original" } }
--
2.35.3


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

end of thread, other threads:[~2022-08-22  9:52 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-08-20 19:11 [PATCH] Fortran: fix simplification of intrinsics IBCLR and IBSET [PR106557] Harald Anlauf
2022-08-22  9:52 ` Mikael Morin

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).