From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([140.186.70.92]:36364)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1QOt2l-0002lO-QI
	for qemu-devel@nongnu.org; Tue, 24 May 2011 10:57:12 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1QOt2k-0000T5-Mw
	for qemu-devel@nongnu.org; Tue, 24 May 2011 10:57:11 -0400
Received: from mail-pz0-f45.google.com ([209.85.210.45]:46745)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1QOt2k-0000Sg-Hq
	for qemu-devel@nongnu.org; Tue, 24 May 2011 10:57:10 -0400
Received: by pzk30 with SMTP id 30so3650329pzk.4
	for <qemu-devel@nongnu.org>; Tue, 24 May 2011 07:57:09 -0700 (PDT)
Sender: Richard Henderson <rth7680@gmail.com>
Message-ID: <4DDBC741.3090607@twiddle.net>
Date: Tue, 24 May 2011 07:57:05 -0700
From: Richard Henderson <rth@twiddle.net>
MIME-Version: 1.0
References: <1305671572-5899-1-git-send-email-jcmvbkbc@gmail.com>
	<201105230012.35944.jcmvbkbc@gmail.com> <4DDA664E.108@twiddle.net>
	<201105240320.36147.jcmvbkbc@gmail.com>
In-Reply-To: <201105240320.36147.jcmvbkbc@gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [PATCH 21/26] target-xtensa: implement unaligned
 exception option
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: Max Filippov <jcmvbkbc@gmail.com>
Cc: qemu-devel@nongnu.org

On 05/23/2011 04:20 PM, Max Filippov wrote:
> I probably just don't get what you call 'out-of-line'. In fact
> do_unaligned_access will be called for every unaligned access, and
> alignment condition will be checked for every access. It just happens
> in other place. Does it have more chances to be optimized better than
> TCG code, or is it less TCG code itself that makes difference?

Out of line meaning not inside TCG code at all.

The fast-path through a tcg_qemu_ld operation handles aligned memory
accesses that hit the TLB.  Unaligned accesses or TLB misses go to
the slow path, in __ld[bwlq]_mmu.  It is in these functions that we
test for ALIGNED_ONLY, which enables the do_unaligned_access hook.


r~