From mboxrd@z Thu Jan 1 00:00:00 1970 Received: with ECARTIS (v1.0.0; list linux-mips); Wed, 28 Jun 2017 10:34:21 +0200 (CEST) Received: from localhost.localdomain ([127.0.0.1]:38118 "EHLO linux-mips.org" rhost-flags-OK-OK-OK-FAIL) by eddie.linux-mips.org with ESMTP id S23993972AbdF1IeNcyc4X (ORCPT ); Wed, 28 Jun 2017 10:34:13 +0200 Received: from h7.dl5rb.org.uk (localhost [127.0.0.1]) by h7.dl5rb.org.uk (8.15.2/8.14.8) with ESMTP id v5S8YCde019683; Wed, 28 Jun 2017 10:34:12 +0200 Received: (from ralf@localhost) by h7.dl5rb.org.uk (8.15.2/8.15.2/Submit) id v5S8YBDq019681; Wed, 28 Jun 2017 10:34:11 +0200 Date: Wed, 28 Jun 2017 10:34:11 +0200 From: Ralf Baechle To: "Maciej W. Rozycki" Cc: Leonid Yegoshin , Matt Redfearn , Aleksandar Markovic , "linux-mips@linux-mips.org" , James Hogan , Paul Burton , Raghu Gandham , Douglas Leung , Petar Jovanovic , Miodrag Dinic , Goran Ferenc Subject: Re: [PATCH 3/8] MIPS: R6: Fix PREF instruction usage by memcpy for MIPS R6 Message-ID: <20170628083411.GD6738@linux-mips.org> References: <1497887415-13825-1-git-send-email-aleksandar.markovic@rt-rk.com> <1497887415-13825-4-git-send-email-aleksandar.markovic@rt-rk.com> <20170628005853.GC6738@linux-mips.org> <9915e476-70fe-4281-a1e4-85ca2e8683b3@imgtec.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.8.0 (2017-02-23) Return-Path: X-Envelope-To: <"|/home/ecartis/ecartis -s linux-mips"> (uid 0) X-Orcpt: rfc822;linux-mips@linux-mips.org Original-Recipient: rfc822;linux-mips@linux-mips.org X-archive-position: 58844 X-ecartis-version: Ecartis v1.0.0 Sender: linux-mips-bounce@linux-mips.org Errors-to: linux-mips-bounce@linux-mips.org X-original-sender: ralf@linux-mips.org Precedence: bulk List-help: List-unsubscribe: List-software: Ecartis version 1.0.0 List-Id: linux-mips X-List-ID: linux-mips List-subscribe: List-owner: List-post: List-archive: X-list: linux-mips On Wed, Jun 28, 2017 at 03:36:54AM +0100, Maciej W. Rozycki wrote: > On Wed, 28 Jun 2017, Leonid Yegoshin wrote: > > > The bigger problem is a prefetch overrun to device registers. AT use is just > > nuisance. > > That isn't however what the patch addresses, not at least according to > the description provided. > > How far beyond the data copied do we prefetch anyway, and can't we simply > waste a page's worth of space or so at the physical end of each RAM area > present in a given system? Or perhaps use it for something we know for > sure that won't ever be copied. Prefetching may also bring data which has just been flushed back into caches so until this is fixed properly it's not a good idea on non-coherent platforms. Ralf