From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43)
	id 1LA89U-0003c8-3C
	for qemu-devel@nongnu.org; Tue, 09 Dec 2008 14:21:48 -0500
Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43)
	id 1LA89S-0003bO-LC
	for qemu-devel@nongnu.org; Tue, 09 Dec 2008 14:21:46 -0500
Received: from [199.232.76.173] (port=38559 helo=monty-python.gnu.org)
	by lists.gnu.org with esmtp (Exim 4.43) id 1LA89S-0003bL-EI
	for qemu-devel@nongnu.org; Tue, 09 Dec 2008 14:21:46 -0500
Received: from mail-bw0-f21.google.com ([209.85.218.21]:63161)
	by monty-python.gnu.org with esmtp (Exim 4.60)
	(envelope-from <laurent.desnogues@gmail.com>) id 1LA89S-0001D4-8p
	for qemu-devel@nongnu.org; Tue, 09 Dec 2008 14:21:46 -0500
Received: by bwz14 with SMTP id 14so195539bwz.10
	for <qemu-devel@nongnu.org>; Tue, 09 Dec 2008 11:21:43 -0800 (PST)
Message-ID: <761ea48b0812091121j2b2526d8gfafd3ac48e379b18@mail.gmail.com>
Date: Tue, 9 Dec 2008 20:21:43 +0100
From: "Laurent Desnogues" <laurent.desnogues@gmail.com>
Subject: Re: [Qemu-devel] [ARM] Problem in a NEON instruction
In-Reply-To: <20081014092338.GA13141@volta.aurel32.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
References: <761ea48b0806100332r16fc2eafxa2e9db5e5afadeb4@mail.gmail.com>
	<20081014092338.GA13141@volta.aurel32.net>
Reply-To: qemu-devel@nongnu.org
List-Id: qemu-devel.nongnu.org
List-Unsubscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.gnu.org/pipermail/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: qemu-devel@nongnu.org

On Tue, Oct 14, 2008 at 10:23 AM, Aurelien Jarno <aurelien@aurel32.net> wrote:
> On Tue, Jun 10, 2008 at 12:32:17PM +0200, Laurent Desnogues wrote:
>> The rshl instruction is faulty.  Note I did not check the correctness
>> of that instruction in general, I only made a change that looked logical.
>>
>>
>> Laurent
>
>> --- svn-ref/target-arm/neon_helper.c  2008-06-09 08:52:48.000000000 +0200
>> +++ svn/target-arm/neon_helper.c      2008-06-10 12:27:38.000000000 +0200
>> @@ -456,7 +456,7 @@
>>      if (tmp >= sizeof(src1) * 8) { \
>>          dest = 0; \
>>      } else if (tmp < -sizeof(src1) * 8) { \
>> -        dest >>= sizeof(src1) * 8 - 1; \
>> +        dest = src1 >> (sizeof(src1) * 8 - 1); \
>>      } else if (tmp == -sizeof(src1) * 8) { \
>>          dest = src1 >> (tmp - 1); \
>>          dest++; \
>
> Your patch is indeed correct. It seems there is another mistake (at
> least compared to the 64-bit version) in this instruction, please find
> an updated patch below.
>
> diff --git a/target-arm/neon_helper.c b/target-arm/neon_helper.c
> index 4ee5658..35fbaf5 100644
> --- a/target-arm/neon_helper.c
> +++ b/target-arm/neon_helper.c
> @@ -456,11 +456,11 @@ uint64_t HELPER(neon_shl_s64)(uint64_t valop, uint64_t shiftop)
>     if (tmp >= sizeof(src1) * 8) { \
>         dest = 0; \
>     } else if (tmp < -sizeof(src1) * 8) { \
> -        dest >>= sizeof(src1) * 8 - 1; \
> +        dest = src1 >> (sizeof(src1) * 8 - 1); \
>     } else if (tmp == -sizeof(src1) * 8) { \
>         dest = src1 >> (tmp - 1); \
>         dest++; \
> -        src2 >>= 1; \
> +        dest >>= 1; \
>     } else if (tmp < 0) { \
>         dest = (src1 + (1 << (-1 - tmp))) >> -tmp; \
>     } else { \

Can someone consider that patch especially now that gcc 4
displays some well deserved warnings about the code :-)


Laurent