From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:36791)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <tommusta@gmail.com>) id 1VtKyq-0000a3-PK
	for qemu-devel@nongnu.org; Wed, 18 Dec 2013 12:32:29 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <tommusta@gmail.com>) id 1VtKyi-0002sM-Bn
	for qemu-devel@nongnu.org; Wed, 18 Dec 2013 12:32:20 -0500
Message-ID: <52B1DC18.4080106@gmail.com>
Date: Wed, 18 Dec 2013 11:32:08 -0600
From: Tom Musta <tommusta@gmail.com>
MIME-Version: 1.0
References: <1386789398-5239-1-git-send-email-tommusta@gmail.com>
	<1386789398-5239-12-git-send-email-tommusta@gmail.com>
	<CAFEAcA_rY49bSr=GnC7rcjER6_WhkVG5WuwTR753xp11eZqMLQ@mail.gmail.com>
	<52A8CD67.8060804@gmail.com>
	<CAFEAcA8T2PZmqDrmXP7EwDOsAEJX+TFNodg9pYKQDw23BF4HxA@mail.gmail.com>
In-Reply-To: <CAFEAcA8T2PZmqDrmXP7EwDOsAEJX+TFNodg9pYKQDw23BF4HxA@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [V2 PATCH 11/18] softfloat: Fix float64_to_uint32
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Peter Maydell <peter.maydell@linaro.org>
Cc: "qemu-ppc@nongnu.org" <qemu-ppc@nongnu.org>, QEMU Developers <qemu-devel@nongnu.org>

On 12/17/2013 11:45 AM, Peter Maydell wrote:
> On 11 December 2013 20:39, Tom Musta <tommusta@gmail.com> wrote:
>> On 12/11/2013 1:53 PM, Peter Maydell wrote:
>>> On 11 December 2013 19:16, Tom Musta <tommusta@gmail.com> wrote:
>>>>  uint32 float64_to_uint32( float64 a STATUS_PARAM )
>>>>  {
>>>> -    int64_t v;
>>>> +    uint64_t v;
>>>>      uint32 res;
>>>>
>>>> -    v = float64_to_int64(a STATUS_VAR);
>>>> -    if (v < 0) {
>>>> -        res = 0;
>>>> -        float_raise( float_flag_invalid STATUS_VAR);
>>>> -    } else if (v > 0xffffffff) {
>>>> +    v = float64_to_uint64(a STATUS_VAR);
>>>> +    if (v > 0xffffffff) {
>>>>          res = 0xffffffff;
>>>> +        STATUS(float_exception_flags) &= ~float_flag_inexact;
>>>
>>> The IEEE exception flags are cumulative (ie never get cleared
>>> except by guest program explicit request); this change means
>>> that if a previous operation raised the inexact flag you've just
>>> lost that.
>>>
>>> thanks
>>> -- PMM
>>>
>> Thank you, Peter.  I will fix.
> 
> I'm partway through fixing this bug in an implementation of
> float*_to_uint16 which the ARM AArch64 needs. I think the
> cleanest approach to this looks like this:
> 
> uint32 float64_to_uint32( float64 a STATUS_PARAM )
> {
>     int64_t v;
>     uint32 res;
>     int old_exc_flags = get_float_exception_flags(status);
> 
>     v = float64_to_uint64(a STATUS_VAR);
>     if (v > 0xffffffff) {
>         res = 0xffffffff;
>     } else {
>         return v;
>     }
>     set_float_exception_flags(old_exc_flags);
>     float_raise(float_flag_invalid STATUS_VAR);
>     return res;
> }
> 
> thanks
> -- PMM
> 

This seems to assume that the only case where flags could be set in
float64_to_uint32 is the case where a large result is returned. Is
this really the case?

I was thinking something like this:

uint32 float64_to_uint32( float64 a STATUS_PARAM )
{
    uint64_t v;
    uint32 res;
    int inexact = (STATUS(float_exception_flags) & float_flag_inexact) != 0;

    v = float64_to_uint64(a STATUS_VAR);
    if (v > 0xffffffff) {
        res = 0xffffffff;
        /* If the inexact flag was set during this operation, then clear it. */
        if (!inexact) {
            STATUS(float_exception_flags) &= ~float_flag_inexact;
        }
        float_raise( float_flag_invalid STATUS_VAR);
    } else {
        res = v;
    }
    return res;
}