Date: Mon, 24 Nov 1997 13:20:27 +1100 (EST) From: Andrew Cagney <cagney@tpgi.com.au> To: FreeBSD-gnats-submit@FreeBSD.ORG Subject: bin/5135: scanf accepts %Lg but treates it as %g Message-ID: <199711240220.NAA18459@b1.tpgi.com.au> Resent-Message-ID: <199711240630.WAA08423@hub.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 5135
>Category: bin
>Synopsis: scanf accepts %Lg but treates it as %g
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: freebsd-bugs
>State: open
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Sun Nov 23 22:30:01 PST 1997
>Last-Modified:
>Originator: Andrew Cagney
>Organization:
>Release: FreeBSD 2.2.5-RELEASE i386
>Environment:
Vanilla 2.2.5
>Description:
Burried away in the scanf documentation is:
L Indicates that the conversion will be efg and the next pointer is
a pointer to long double. (This type is not implemented; the L
flag is currently ignored.)
to be more precise, it assigns a `%g' value (32 bit float or single)
to a `%Lg' destination (80 bit long double). I'm sorry but doing this
is worse than stupid!
>How-To-Repeat:
One example, all GDB floating point is broken:
$ gdb
...
(gdb) print 1.0
$1 = 0
If you trace it through, c-exp.c:parse_number calls sscanf(...%Lg...).
>Fix:
For FreeBSD:
o modify scanf so that for, %Lg (long double) it parses
it as %lg (double) then extends/stores it as a long double.
o Remove all support for %Lg in both printf and scanf.
Some software (eg gdb) tests printf ("%Lg") and then assumes
scanf also works. Consequently it is an all or nothing afair
:-)
For GDB:
I'll see if I can get GDB changed so that its configure script
separately checks for support of %Lg in scanf and printf.
Have a nice day.
Andrew
>Audit-Trail:
>Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711240220.NAA18459>