From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 21884 invoked by alias); 8 Jan 2007 12:43:02 -0000 Received: (qmail 21876 invoked by uid 22791); 8 Jan 2007 12:43:02 -0000 X-Spam-Check-By: sourceware.org Received: from province.act-europe.fr (HELO province.act-europe.fr) (212.157.227.214) by sourceware.org (qpsmtpd/0.31) with ESMTP; Mon, 08 Jan 2007 12:42:55 +0000 Received: from localhost (localhost [127.0.0.1]) by filtered-province.act-europe.fr (Postfix) with ESMTP id 889B24AD06; Mon, 8 Jan 2007 13:42:53 +0100 (CET) Received: from province.act-europe.fr ([127.0.0.1]) by localhost (province.act-europe.fr [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 17791-03; Mon, 8 Jan 2007 13:42:53 +0100 (CET) Received: from [192.168.1.3] (dyn-88-122-77-190.ppp.tiscali.fr [88.122.77.190]) (using TLSv1 with cipher RC4-MD5 (128/128 bits)) (No client certificate requested) by province.act-europe.fr (Postfix) with ESMTP id 433CD4ACEF; Mon, 8 Jan 2007 13:42:53 +0100 (CET) From: Eric Botcazou To: "Richard Guenther" Subject: Re: [PATCH] Tree SRA and atomicity/volatility Date: Mon, 08 Jan 2007 12:43:00 -0000 User-Agent: KMail/1.7.1 Cc: gcc-patches@gcc.gnu.org References: <200701061422.39157.ebotcazou@adacore.com> <200701081233.10301.ebotcazou@adacore.com> <84fc9c000701080352x517b891cva220bc67f75a2a06@mail.gmail.com> In-Reply-To: <84fc9c000701080352x517b891cva220bc67f75a2a06@mail.gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200701081345.53442.ebotcazou@adacore.com> Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org X-SW-Source: 2007-01/txt/msg00568.txt.bz2 > Oh, so the only problem you see are for structure copies that are > decomposed by SRA? At the moment, yes. > Still, middle-end support for this kind of stuff could be useful - maybe > using volatile bitfields instead of a volatile structure type. > > struct val { > volatile unsigned val1 : 8; > volatile unsigned val2 : 8; > }; > > (at the moment accesses to the above are also decomposed by expand). Why would we prevent the decomposition in this case? -- Eric Botcazou