From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753883AbcAORGy (ORCPT ); Fri, 15 Jan 2016 12:06:54 -0500 Received: from mx1.redhat.com ([209.132.183.28]:57045 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751575AbcAORGx (ORCPT ); Fri, 15 Jan 2016 12:06:53 -0500 Subject: Re: kvm: access to invalid memory in mmu_zap_unsync_children To: Sasha Levin , Gleb Natapov References: <56992438.6090804@oracle.com> Cc: LKML , Dmitry Vyukov , syzkaller , Marcelo Tosatti , Takuya Yoshikawa , Xiao Guangrong From: Paolo Bonzini Message-ID: <56992719.9060004@redhat.com> Date: Fri, 15 Jan 2016 18:06:33 +0100 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.4.0 MIME-Version: 1.0 In-Reply-To: <56992438.6090804@oracle.com> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 15/01/2016 17:54, Sasha Levin wrote: > Hi all, > > While fuzzing with syzkaller on the latest -next kernel running on a KVM tools > guest, I've hit the following invalid memory access: > > [ 547.956284] UBSAN: Undefined behaviour in arch/x86/kvm/mmu.c:2011:17 > > [ 547.956940] index 3 is out of range for type 'kvm_mmu_page *[3]' > > [ 547.957567] CPU: 0 PID: 21577 Comm: syz-executor Tainted: G D 4.4.0-next-20160114-sasha-00021-gf1273d1-dirty #2798 > > [ 547.958739] 1ffff1001819be5c 000000002fa0e55b ffff8800c0cdf360 ffffffff83433c4e > > [ 547.972448] 0000000041b58ab3 ffffffff8f960c38 ffffffff83433b86 ffff8800c0cdf328 > > [ 547.973277] 0000000000000001 000000002fa0e55b ffffffff8feb8440 ffff8800c0cdf3f0 > > [ 547.974102] Call Trace: > > [ 547.974424] dump_stack (lib/dump_stack.c:52) > [ 547.975774] ubsan_epilogue (lib/ubsan.c:165) > [ 547.976408] __ubsan_handle_out_of_bounds (lib/ubsan.c:382) > [ 547.980877] mmu_zap_unsync_children (arch/x86/kvm/mmu.c:2011 arch/x86/kvm/mmu.c:2272) Marcelo/Takuya/Xiao, do you know what's the point in the assignment in kvm_mmu_pages_init? It seems to me that it should be parents->parent[0] = NULL; since the only user of the ->parent[] array, mmu_pages_clear_parents, walks the array up from parents->parent[0]. Any other opinions? Paolo