From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Message-ID: <51BA2FF4.6000604@zytor.com> Date: Thu, 13 Jun 2013 13:47:48 -0700 From: "H. Peter Anvin" MIME-Version: 1.0 To: Yinghai Lu CC: Thomas Gleixner , Ingo Molnar , linux-kernel@vger.kernel.org, stable@vger.kernel.org Subject: Re: [PATCH] x86: Fix /proc/mtrr with base/size more than 44bits References: <1371149617-17880-1-git-send-email-yinghai@kernel.org> In-Reply-To: <1371149617-17880-1-git-send-email-yinghai@kernel.org> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: On 06/13/2013 11:53 AM, Yinghai Lu wrote: > > - if (base & size_or_mask || size & size_or_mask) { > + if (base >> (boot_cpu_data.x86_phys_bits - PAGE_SHIFT) || > + base > (base + size) || > + (base + size - 1) >> (boot_cpu_data.x86_phys_bits - PAGE_SHIFT)) { > pr_warning("mtrr: base or size exceeds the MTRR width\n"); > return -EINVAL; > } Most of this patch looks good as far as being a minimal patch, but I'm really confused about this bit. Could you explain the reason for why the original doesn't work? (To be fair: I am not even sure the original does anything useful so it could just be a "this is just too broken to live" kind of thing.) The first and third clause of the test can be simplified, however: (base | (base + size - 1)) >> (boot_cpu_data.x86_phys_bits - PAGE_SHIFT) ... although it would be cleaner to put boot_cpu_data.x86_phys_bits - PAGE_SHIFT into a variable. A lot of the mask_hi/mask_lo stuff should just get removed by using rdmsrl/wrmsrl, but that is not stable material obviously. -hpa