From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nick Piggin Subject: Re: MMIO and gcc re-ordering issue Date: Wed, 4 Jun 2008 12:19:27 +1000 Message-ID: <200806041219.28099.nickpiggin@yahoo.com.au> References: <1211852026.3286.36.camel@pasglop> <200806031416.18195.nickpiggin@yahoo.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from smtp111.mail.mud.yahoo.com ([209.191.84.64]:46118 "HELO smtp111.mail.mud.yahoo.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1751311AbYFDCTp (ORCPT ); Tue, 3 Jun 2008 22:19:45 -0400 In-Reply-To: Content-Disposition: inline Sender: linux-arch-owner@vger.kernel.org List-ID: To: Linus Torvalds Cc: Russell King , Benjamin Herrenschmidt , David Miller , linux-arch@vger.kernel.org, scottwood@freescale.com, linuxppc-dev@ozlabs.org, alan@lxorguk.ukuu.org.uk, linux-kernel@vger.kernel.org, tpiepho@freescale.com On Wednesday 04 June 2008 00:47, Linus Torvalds wrote: > On Tue, 3 Jun 2008, Nick Piggin wrote: > > Linus: on x86, memory operations to wc and wc+ memory are not ordered > > with one another, or operations to other memory types (ie. load/load > > and store/store reordering is allowed). Also, as you know, store/load > > reordering is explicitly allowed as well, which covers all memory > > types. So perhaps it is not quite true to say readl/writel is strongly > > ordered by default even on x86. You would have to put in some > > mfence instructions in them to make it so. > > Well, you have to ask for WC/WC+ anyway, so it's immaterial. A driver that > does that needs to be aware of it. IOW, it's a non-issue, imnsho. Ah, yes UC is strongly ordered WRT all others *except* WC/WC+. But WC memory is not an x86 specific thing right, so do we need some accessors for WC memory? Or can we just throw that in the weakly ordered pile, and ensure mb/rmb/wmb does the right thing for them. And you want readl/writel to be strongly ordered like x86 on all architectures, no exceptions? This will slow some things down, but if we then also provide explicitly weakly ordered instructions (and add io_mb/io_rmb/io_wmb) then at least it gives the framework for drivers to be written to run on those architectures. The other thing we could do is mandate only that readl/writel will be ordered WRT one another, *and* with spinlocks, but otherwise not with cacheable RAM...