From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934809Ab1JEOtr (ORCPT ); Wed, 5 Oct 2011 10:49:47 -0400 Received: from mail-ww0-f44.google.com ([74.125.82.44]:57448 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753471Ab1JEOtp (ORCPT ); Wed, 5 Oct 2011 10:49:45 -0400 MIME-Version: 1.0 In-Reply-To: <1317798113.29415.270.camel@pasglop> References: <1317653420-21404-3-git-send-email-mason@myri.com> <1317742838.29415.217.camel@pasglop> <1317744493.29415.238.camel@pasglop> <1317749439.29415.251.camel@pasglop> <1317798113.29415.270.camel@pasglop> From: Linus Torvalds Date: Wed, 5 Oct 2011 07:49:24 -0700 X-Google-Sender-Auth: 1H6NnItrtdU1vC0GVyH5-ztAMEY Message-ID: Subject: Re: [PATCH 2/3] pci: Clamp pcie_set_readrq() when using "performance" settings To: Benjamin Herrenschmidt Cc: Jon Mason , Greg Kroah-Hartman , Jesse Barnes , Bjorn Helgaas , linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org, Benjamin LaHaise Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Oct 5, 2011 at 12:01 AM, Benjamin Herrenschmidt wrote: > >> We just don't the hell know, do we? > > Well, we do with some confidence :-) Or rather what we do know is what > you have today in your tree is broken. You're missing the point. Repeat after me: late -rc series is not when we test these kinds of things. This point in time is when we *revert* commits that are broken. We fix them if there is absolutely no question about the fix, but that simply isn't true here. Even if the patches "improve" something, there is no way I hell that I believe that we suddenly don't need to worry about MPS any more. How hard is this to just understand? It's not about "we can improve things". It's about "it's f*%!ing late in the rc series, we're not dicking around any more!". So quite frankly, if you don't like the code now, send me a revert for all the mess. BUT DON'T ARGUE FOR CHANGES THAT AREN'T 100% ROCK SOLID. So the current situation is that MPS is simply *disabled*. All the crap code simply doesn't matter at all, because nobody will run it. Arguing that it is "broken" is stupid, because the only people that that brokenness would ever matter for are the people like you who are testing things out - not actual users. Comprende? Think of it like a stable kernel. We don't mess around with things that don't matter and nobody will hit. We don't do "development" in the late -rc, and yet that is what the MPS patches in question have been doing. IT IS TOO LATE FOR CRAP LIKE THAT. WE TURNED IT OFF. Linus