From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753476AbaIXQHS (ORCPT ); Wed, 24 Sep 2014 12:07:18 -0400 Received: from outbound-smtp02.blacknight.com ([81.17.249.8]:35948 "EHLO outbound-smtp02.blacknight.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751989AbaIXQHP (ORCPT ); Wed, 24 Sep 2014 12:07:15 -0400 Message-ID: <5422EC30.10206@nexus-software.ie> Date: Wed, 24 Sep 2014 17:07:12 +0100 From: "Bryan O'Donoghue" User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.1.1 MIME-Version: 1.0 To: "H. Peter Anvin" , mingo@kernel.org, linux-kernel@vger.kernel.org, tglx@linutronix.de, stable@vger.kernel.org, bp@alien8.de, linux-tip-commits@vger.kernel.org Subject: Re: [tip:x86/platform] x86/intel/quark: Switch off CR4.PGE so TLB flush uses CR3 instead References: <1411514784-14885-1-git-send-email-pure.logic@nexus-software.ie> <5422E8D0.5080705@zytor.com> <5422EA13.1080706@nexus-software.ie> In-Reply-To: <5422EA13.1080706@nexus-software.ie> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 24/09/14 16:58, Bryan O'Donoghue wrote: > >> >> I believe there is one more change needed: there is a __flush_tlb_all() >> in the early code which gets executed before the above code runs; the >> easiest fix is to just add a __flush_tlb() immediately after it. >> >> This should have been pushed upstream, and not stayed in the BSP kernel. > > Peter. > > You're talking about void __init setup_arch() right ? > > The code looks like this > > load_cr3(swapper_pg_dir); > __flush_tlb_all(); > > Note that on Quark the method to invalidate the TLB is by reloading CR3 > - which is immediately prior to __flush_tlb_all(); > > So __flush_tlb_all(); will do nothing but that's OK since load_cr3() > already flushed the TLB - including any PTE with PGE set I take your point though. It's probably better to be explicit and do the whole if (quark) __flush_tlb(); else __flush_tlb_all(); So that we aren't relying on the side effects of the previous statement.