NetBSD Problem Report #47536

From martin@aprisoft.de  Wed Feb  6 14:32:52 2013
Return-Path: <martin@aprisoft.de>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
	by www.NetBSD.org (Postfix) with ESMTP id 38E8D63E8B1
	for <gnats-bugs@gnats.NetBSD.org>; Wed,  6 Feb 2013 14:32:52 +0000 (UTC)
Message-Id: <20130206143245.452CAED0E4C@emmas.aprisoft.de>
Date: Wed,  6 Feb 2013 15:32:45 +0100 (CET)
From: martin@NetBSD.org
Reply-To: martin@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: printf conversion fails for long double on sparc64
X-Send-Pr-Version: 3.95

>Number:         47536
>Category:       port-sparc64
>Synopsis:       printf conversion fails for long double on sparc64
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    martin
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Feb 06 14:35:01 +0000 2013
>Closed-Date:    Sat Jul 13 20:14:06 +0000 2013
>Last-Modified:  Sat Jul 13 20:14:06 +0000 2013
>Originator:     Martin Husemann
>Release:        NetBSD 6.99.16
>Organization:
The NetBSD Foundation, Inc.
>Environment:
System: NetBSD whoever-brings-the-night.aprisoft.de 6.99.16 NetBSD 6.99.16 (WHOEVER) #20: Mon Feb 4 12:20:15 CET 2013 martin@emmas.aprisoft.de:/usr/src/sys/arch/sparc64/compile/WHOEVER sparc64
Architecture: sparc64
Machine: sparc64
>Description:

This is a test program by Edward Berner posted to the Fossil users list:

#include <stdio.h>

int
main (int argc, char *argv[])
{
	long long n;
	long double ld;

	n = 2147483648LL;
	ld = n; 
	printf (" n = %lld\n", n);
	printf ("ld = %0.17Lg\n", ld);

	return 0;
}


When run on sparc64 it prints:
 n = 2147483648
ld = 0


>How-To-Repeat:
s/a

>Fix:
n/a

>Release-Note:

>Audit-Trail:

Responsible-Changed-From-To: port-sparc64-maintainer->martin
Responsible-Changed-By: martin@NetBSD.org
Responsible-Changed-When: Wed, 06 Feb 2013 14:37:14 +0000
Responsible-Changed-Why:
Take


From: David Laight <david@l8s.co.uk>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-sparc64/47536: printf conversion fails for long double on sparc64
Date: Wed, 6 Feb 2013 21:36:47 +0000

 On Wed, Feb 06, 2013 at 02:35:01PM +0000, martin@NetBSD.org wrote:
 > >Number:         47536
 > >Category:       port-sparc64
 > >Synopsis:       printf conversion fails for long double on sparc64
 ...
 > 	long long n;
 > 	long double ld;
 > 
 > 	n = 2147483648LL;
 > 	ld = n; 
 > 	printf (" n = %lld\n", n);
 > 	printf ("ld = %0.17Lg\n", ld);
 ...
 > When run on sparc64 it prints:
 >  n = 2147483648
 > ld = 0

 Is that a printf() problem or a 'long long' to 'long double'
 convertion problem?

 	David

 -- 
 David Laight: david@l8s.co.uk

From: Jukka Ruohonen <jruohonen@iki.fi>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-sparc64/47536: printf conversion fails for long double on sparc64
Date: Thu, 7 Feb 2013 17:12:27 +0200

 On Wed, Feb 06, 2013 at 02:35:01PM +0000, martin@NetBSD.org wrote:
 > This is a test program by Edward Berner posted to the Fossil users list:
 > 
 > #include <stdio.h>
 > 
 > int
 > main (int argc, char *argv[])
 > {
 > 	long long n;
 > 	long double ld;
 > 
 > 	n = 2147483648LL;
 > 	ld = n; 
 > 	printf (" n = %lld\n", n);
 > 	printf ("ld = %0.17Lg\n", ld);
 > 
 > 	return 0;
 > }

 Looks like a test case? Perhaps if augmented with some other comparable
 cases, similar problems in other architectures could be also revealed
 (cf. e.g. the innocent PR lib/22019; lib/32951; lib/44113).

 - Jukka.

From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-sparc64/47536: printf conversion fails for long double on sparc64
Date: Wed, 13 Feb 2013 08:30:10 +0100

 [forwarded to gnats to update the ticket]

 Martin

 ---8<---

 From: christos@astron.com (Christos Zoulas)
 Subject: Re: port-sparc64/47536: printf conversion fails for long double on sparc64
 Date: Wed, 13 Feb 2013 03:31:17 +0000 (UTC)


 In article <20130207154009.746AE63EF83@www.NetBSD.org>,
 Jukka Ruohonen  <gnats-bugs@NetBSD.org> wrote:
 >The following reply was made to PR port-sparc64/47536; it has been noted
 >by GNATS.
 >
 >From: Jukka Ruohonen <jruohonen@iki.fi>
 >To: gnats-bugs@NetBSD.org
 >Cc: 
 >Subject: Re: port-sparc64/47536: printf conversion fails for long double
 >on sparc64
 >Date: Thu, 7 Feb 2013 17:12:27 +0200
 >
 > On Wed, Feb 06, 2013 at 02:35:01PM +0000, martin@NetBSD.org wrote:
 > > This is a test program by Edward Berner posted to the Fossil users list:
 > > 
 > > #include <stdio.h>
 > > 
 > > int
 > > main (int argc, char *argv[])
 > > {
 > > 	long long n;
 > > 	long double ld;
 > > 
 > > 	n = 2147483648LL;
 > > 	ld = n; 
 > > 	printf (" n = %lld\n", n);
 > > 	printf ("ld = %0.17Lg\n", ld);
 > > 
 > > 	return 0;
 > > }
 > 
 > Looks like a test case? Perhaps if augmented with some other comparable
 > cases, similar problems in other architectures could be also revealed
 > (cf. e.g. the innocent PR lib/22019; lib/32951; lib/44113).

 The sign bit is not set in the long long -> long double conversion.
 Printf is not at fault.

 #include <stdio.h>
 int
 main(int argc, char *argv[])
 {
         union {
                 long double ld;
                 long long n[2];
         } u;

         u.ld = 2147483648LL;
 #ifdef __sparc64__
         u.n[1] = 0x0000000080000000llu;
 #endif
         printf("%zu, %zu\n", sizeof(u.ld), sizeof(u.n));
         printf(" u.n %llx %llx\n", u.n[0], u.n[1]);
         printf("u.ld %0.17Lg\n", u.ld);
         return 0;
 }

 christos

From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-sparc64/47536: printf conversion fails for long double on sparc64
Date: Wed, 13 Feb 2013 09:53:00 +0100

 The gcc generated code for the call site is correct. For the original test
 program it placess the format string pointer in %o0, and the correct long
 double value (pre generated at compile time) in %o2 and %o3.

 Sounds like a va_list problem with the padding (%o1 being skipped).

 Martin

From: "Martin Husemann" <martin@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/47536 CVS commit: src/sys/arch/sparc/include
Date: Thu, 14 Feb 2013 09:34:46 +0000

 Module Name:	src
 Committed By:	martin
 Date:		Thu Feb 14 09:34:46 UTC 2013

 Modified Files:
 	src/sys/arch/sparc/include: ieee.h

 Log Message:
 We have an implicit 1 MSB bit in the mantissa. Pointed out by Matt Thomas.
 Fixes PR port-sparc64/47536.


 To generate a diff of this commit:
 cvs rdiff -u -r1.14 -r1.15 src/sys/arch/sparc/include/ieee.h

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

State-Changed-From-To: open->pending-pullups
State-Changed-By: martin@NetBSD.org
State-Changed-When: Thu, 14 Feb 2013 09:35:50 +0000
State-Changed-Why:
Fixed in current, needs pullup.


From: "Julian Coleman" <jdc@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/47536 CVS commit: [netbsd-6] src/sys/arch/sparc/include
Date: Thu, 14 Feb 2013 22:12:11 +0000

 Module Name:	src
 Committed By:	jdc
 Date:		Thu Feb 14 22:12:10 UTC 2013

 Modified Files:
 	src/sys/arch/sparc/include [netbsd-6]: ieee.h

 Log Message:
 Pull up revision 1.15 (requested by martin in ticket #821).

  We have an implicit 1 MSB bit in the mantissa. Pointed out by Matt Thomas.
  Fixes PR port-sparc64/47536.


 To generate a diff of this commit:
 cvs rdiff -u -r1.14 -r1.14.8.1 src/sys/arch/sparc/include/ieee.h

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

State-Changed-From-To: pending-pullups->closed
State-Changed-By: martin@NetBSD.org
State-Changed-When: Sat, 13 Jul 2013 20:14:06 +0000
State-Changed-Why:
Pullups done


>Unformatted:

Home	PR Database Search