From: Arnaud Charlet <charlet@ACT-Europe.FR>
To: gcc-patches@gcc.gnu.org
Cc: Ed Schonberg <schonberg@gnat.com>
Subject: Ada: fix validity check on some floating point operation
Date: Wed, 27 Oct 2004 12:12:00 -0000 [thread overview]
Message-ID: <20041027102052.GJ91230@province.act-europe.fr> (raw)
[-- Attachment #1: Type: text/plain, Size: 853 bytes --]
Tested on x86-linux
Validity checks were properly enabled on the operands of these operations,
but not on the result, which allowed the generation and propagation of abnormal
values (Infinities and NaN's).
Test p_test_float verifies that invalid values are properly
detected:
with TEXT_IO;
use TEXT_IO;
procedure P_TEST_FLOAT is
V_1, V_2, V_3 : Long_Float;
begin
V_1 := 0.004389773266958945;
V_2 := 0.0;
V_3 := V_1 / V_2;
put_line ("Failed: division by zero");
exception
when constraint_error => put_line ("division by zero detected");
end P_TEST_FLOAT;
$ gnatmake -q -gnatVirofc p_test_float
2004-10-26 Ed Schonberg <schonberg@gnat.com>
* checks.adb (Expr_Known_Valid): If floating-point validity checks are
enabled, check the result of unary and binary operations when the
expression is the right-hand side of an assignment.
[-- Attachment #2: difs.8 --]
[-- Type: text/plain, Size: 1493 bytes --]
Index: checks.adb
===================================================================
RCS file: /cvs/gcc/gcc/gcc/ada/checks.adb,v
retrieving revision 1.29
diff -u -p -r1.29 checks.adb
--- checks.adb 4 Oct 2004 14:58:46 -0000 1.29
+++ checks.adb 27 Oct 2004 09:32:36 -0000
@@ -3792,13 +3792,26 @@ package body Checks is
-- The result of any function call or operator is always considered
-- valid, since we assume the necessary checks are done by the call.
+ -- For operators on floating-point operations, we must also check
+ -- when the operation is the right-hand side of an assignment, or
+ -- is an actual in a call.
- elsif Nkind (Expr) in N_Binary_Op
- or else
- Nkind (Expr) in N_Unary_Op
- or else
- Nkind (Expr) = N_Function_Call
+ elsif
+ Nkind (Expr) in N_Binary_Op or else Nkind (Expr) in N_Unary_Op
then
+ if Is_Floating_Point_Type (Typ)
+ and then Validity_Check_Floating_Point
+ and then
+ (Nkind (Parent (Expr)) = N_Assignment_Statement
+ or else Nkind (Parent (Expr)) = N_Function_Call
+ or else Nkind (Parent (Expr)) = N_Parameter_Association)
+ then
+ return False;
+ else
+ return True;
+ end if;
+
+ elsif Nkind (Expr) = N_Function_Call then
return True;
-- For all other cases, we do not know the expression is valid
next reply other threads:[~2004-10-27 10:22 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-10-27 12:12 Arnaud Charlet [this message]
2004-10-27 12:37 ` Arnaud Charlet
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20041027102052.GJ91230@province.act-europe.fr \
--to=charlet@act-europe.fr \
--cc=gcc-patches@gcc.gnu.org \
--cc=schonberg@gnat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).