From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [140.186.70.92] (port=59612 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1Q6LNT-0000Bk-Nb for qemu-devel@nongnu.org; Sun, 03 Apr 2011 07:22:01 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Q6LNO-00018M-Kr for qemu-devel@nongnu.org; Sun, 03 Apr 2011 07:21:55 -0400 Received: from mail-vx0-f173.google.com ([209.85.220.173]:32768) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Q6LNO-00017o-Ho for qemu-devel@nongnu.org; Sun, 03 Apr 2011 07:21:50 -0400 Received: by vxb41 with SMTP id 41so4138052vxb.4 for ; Sun, 03 Apr 2011 04:21:49 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: References: <1301668243-29886-1-git-send-email-peter.maydell@linaro.org> <1301668243-29886-2-git-send-email-peter.maydell@linaro.org> Date: Sun, 3 Apr 2011 12:21:49 +0100 Message-ID: Subject: Re: [Qemu-devel] [PATCH 01/10] target-arm: Make Neon helper routines use correct FP status From: Peter Maydell Content-Type: text/plain; charset=UTF-8 List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Blue Swirl Cc: Aurelien Jarno , qemu-devel@nongnu.org On 3 April 2011 12:10, Blue Swirl wrote: > On Sun, Apr 3, 2011 at 1:51 PM, Peter Maydell wrote: >> But I can submit a patch to do that if it's the right thing. > > It's not so much about correctness, but performance. All generated > code already has access to global env, so passing it via helper > arguments requires extra instructions which can be avoided. Yes; I meant "right thing" in the more general senses of "is best practice for qemu code" and "causes my patches to be accepted" :-) Anyway, how about I do a version of this patch which moves the affected neon helpers to op_helper.c rather than adding an env parameter (which would avoid changes that a cleanup would just have to revert); but leave patch 10 as-is (that's the one which is touching vfp helpers, but it is just cleanup which isn't changing how they handle env) ? Then I can do a separate patchset to move other helpers, rather than tangling a code-cleanup patchset with Neon correctness fixes. -- PMM