I have been running all of my extensive tests through valgrind and picking
off errors, one by one. I found logic in gcc/real.cc that has an
inconsequential read of an uninitialized memory location. This change
eliminates that warning.
Is the following patch okay for trunk?
>From fffd7c46b2796e5ff98a53b06409f48961e1eb21 Mon Sep 17 00:00:00 2001
From: Robert Dubner <rdub...@symas.com>
Date: Thu, 31 Jul 2025 07:45:26 -0400
Subject: [PATCH] real: Eliminate access to uninitialized memory.
When compiling this program with gcobol:
identification division.
program-id. prog.
data division.
working-storage section.
01 val pic v9(5) value .001.
procedure division.
display val
goback.
the rounding up of .99999...9999 to 1.000...0000 causes a read of the
first byte of the output buffer. Although harmless, it generates a
valgrind warning. The following change clears that warning.
gcc/ChangeLog:
* real.cc (real_to_decimal_for_mode): Set str[0] to known value.
---
gcc/real.cc | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/gcc/real.cc b/gcc/real.cc
index 1f987d48889..43d25246ed7 100644
--- a/gcc/real.cc
+++ b/gcc/real.cc
@@ -1629,6 +1629,11 @@ real_to_decimal_for_mode (char *str, const
REAL_VALUE_TYPE *r_orig,
strcpy (str, (r.sign ? "-0.0" : "0.0"));
return;
case rvc_normal:
+ /* When r_orig is a positive value that converts to all nines and
is
+ rounded up to 1.0, str[0] is harmlessly accessed before being
set to
+ '1'. That read access triggers a valgrind warning. Setting
str[0]
+ to any value quiets the warning. */
+ str[0] = ' ';
break;
case rvc_inf:
strcpy (str, (r.sign ? "-Inf" : "+Inf"));
--
2.34.1
