From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on inbox.vuxu.org
X-Spam-Level: 
X-Spam-Status: No, score=0.2 required=5.0 tests=DKIM_INVALID,DKIM_SIGNED
	autolearn=no autolearn_force=no version=3.4.4
Received: (qmail 20155 invoked from network); 30 Dec 2022 13:21:27 -0000
Received: from 9front.inri.net (168.235.81.73)
  by inbox.vuxu.org with ESMTPUTF8; 30 Dec 2022 13:21:27 -0000
Received: from mail.cosarara.me ([159.69.205.102]) by 9front; Fri Dec 30 08:20:06 -0500 2022
Received: from [192.168.49.141] (unknown [93.176.172.208])
	by mail.cosarara.me (Postfix) with ESMTPSA id 99EF8130CC
	for <9front@9front.org>; Fri, 30 Dec 2022 14:20:00 +0100 (CET)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cosarara.me; s=20200226;
	t=1672406400;
	h=from:from:reply-to:subject:subject:date:date:message-id:message-id:
	 to:to:cc:mime-version:mime-version:content-type:content-type:
	 content-transfer-encoding:content-transfer-encoding;
	bh=yNM8LQf8/R7pD7cwXeGspdBO0f/uEQb/aFfGZIkZD4U=;
	b=i9wK3OgjNw9D1WZDWjJ3RHJ4wnAolVodMpEjWLqqgqxEaLO80XeLzjovrVckQb194JwqAD
	JOlxdWlLc6AbF8vcsfyFJSVQxnjW3diEvvBEHxTpcdpW2nbZ3SzThgGFMByFERsmzGIp0Y
	SlruM0uyXBDkegTBdvLlkV3VapRL4XY=
Message-ID: <6d3b100d-1732-efa2-f01c-728440859ba2@cosarara.me>
Date: Fri, 30 Dec 2022 14:19:59 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.2
From: cosarara <cosa@cosarara.me>
To: 9front@9front.org
Content-Language: en-US
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: quoted-printable
List-ID: <9front.9front.org>
List-Help: <http://lists.9front.org>
X-Glyph: ➈
X-Bullshit: virtual advanced rich-client polling manager 
Subject: [9front] Bug converting vlong to double on 32bit architectures
Reply-To: 9front@9front.org
Precedence: bulk

Hi,

While trying to run kvik's lu9 on ARM, I found that when converting=20
LLONG_MIN to a double on 32bit systems, the result is wrong (positive=20
instead of negative).

Given the following test program:

#include <u.h>
#include <libc.h>

#define LLONG_MAX
#define LLONG_MIN    (-LLONG_MAX-1)

void main() {
     vlong min =3D LLONG_MIN;
     double dmin =3D min;
     print("minint %lld\n", min);
     print("minint as double %f\n", dmin);
     if (dmin > 0.0) {
         exits("int min as double turned positive");
     }
     exits(0);
}

The output on x86_64 will be:

minint -9223372036854775808
minint as double -9223372036854776400.000000

But on arm or 386 (and I expect also spim, 68000, mips, 68020, sparc,=20
power, since they all use the same _v2d):

minint -9223372036854775808
minint as double 9223372036854776400.000000

And the value turned positive in the conversion.

The function used for the cast to double is (in /sys/src/libc/arm/vlrt.c)=
:

#define    SIGN(n)    (1UL<<(n-1))

double
_v2d(Vlong x)
{
     if(x.hi & SIGN(32)) {
         if(x.lo) {
             x.lo =3D -x.lo;
             x.hi =3D ~x.hi;
         } else
             x.hi =3D -x.hi;
         return -((long)x.hi*4294967296. + x.lo);
     }
     return (long)x.hi*4294967296. + x.lo;
}

If I understand correctly, the issue is that where it tries to flip the=20
sign for x.hi (x.hi =3D -x.hi), 0x80000000 has no positive, thus stays th=
e=20
same (it stays negative). Then when we get to the negative return, we=20
get a positive out.

What came to my mind then, is that in the case that there is no x.lo, we=20
can keep the x.hi sign and cast directly, thus:

double
_v2d(Vlong x)
{
 =C2=A0=C2=A0=C2=A0=C2=A0if(!x.lo) {
 =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 return (long)x.hi*4294967296.=
;
 =C2=A0=C2=A0=C2=A0=C2=A0}
 =C2=A0=C2=A0=C2=A0=C2=A0if(x.hi & SIGN(32)) {
 =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 x.lo =3D -x.lo;
 =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 x.hi =3D ~x.hi;
 =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 return -((long)x.hi*429496729=
6. + x.lo);
 =C2=A0=C2=A0=C2=A0=C2=A0}
 =C2=A0=C2=A0=C2=A0=C2=A0return (long)x.hi*4294967296. + x.lo;
}

This looks correct to me, but I don't trust myself to not make mistakes=20
in such critical code, so I would like some feedback on the change.

Happy new year in advance,

cosa