From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pb0-f44.google.com (mail-pb0-f44.google.com [209.85.160.44]) by kanga.kvack.org (Postfix) with ESMTP id 765116B0069 for ; Fri, 6 Jun 2014 10:21:49 -0400 (EDT) Received: by mail-pb0-f44.google.com with SMTP id rq2so2530978pbb.31 for ; Fri, 06 Jun 2014 07:21:49 -0700 (PDT) Received: from mailout2.w1.samsung.com (mailout2.w1.samsung.com. [210.118.77.12]) by mx.google.com with ESMTPS id bz3si19765362pbd.157.2014.06.06.07.21.48 for (version=TLSv1 cipher=RC4-MD5 bits=128/128); Fri, 06 Jun 2014 07:21:48 -0700 (PDT) Received: from eucpsbgm2.samsung.com (unknown [203.254.199.245]) by mailout2.w1.samsung.com (Oracle Communications Messaging Server 7u4-24.01(7.0.4.24.0) 64bit (built Nov 17 2011)) with ESMTP id <0N6R002Z83VXKD00@mailout2.w1.samsung.com> for linux-mm@kvack.org; Fri, 06 Jun 2014 15:21:33 +0100 (BST) Message-id: <5391CD5A.20503@samsung.com> Date: Fri, 06 Jun 2014 18:16:58 +0400 From: Andrey Ryabinin MIME-version: 1.0 Subject: Re: [PATCH] mm: rmap: fix use-after-free in __put_anon_vma References: <1402054255-4930-1-git-send-email-a.ryabinin@samsung.com> <20140606115620.GS3213@twins.programming.kicks-ass.net> In-reply-to: <20140606115620.GS3213@twins.programming.kicks-ass.net> Content-type: text/plain; charset=UTF-8 Content-transfer-encoding: 7bit Sender: owner-linux-mm@kvack.org List-ID: To: Peter Zijlstra Cc: Andrew Morton , Linus Torvalds , linux-mm@kvack.org, linux-kernel@vger.kernel.org, dvyukov@google.com, koct9i@gmail.com On 06/06/14 15:56, Peter Zijlstra wrote: > On Fri, Jun 06, 2014 at 03:30:55PM +0400, Andrey Ryabinin wrote: >> While working address sanitizer for kernel I've discovered use-after-free >> bug in __put_anon_vma. >> For the last anon_vma, anon_vma->root freed before child anon_vma. >> Later in anon_vma_free(anon_vma) we are referencing to already freed anon_vma->root >> to check rwsem. >> This patch puts freeing of child anon_vma before freeing of anon_vma->root. > > Yes, I think that is right indeed. > > Very hard to hit, but valid since not all callers hold rcu_read_lock(). > >> >> Cc: stable@vger.kernel.org # v3.0+ >> Signed-off-by: Andrey Ryabinin >> --- >> mm/rmap.c | 7 ++++--- >> 1 file changed, 4 insertions(+), 3 deletions(-) >> >> diff --git a/mm/rmap.c b/mm/rmap.c >> index 9c3e773..161bffc7 100644 >> --- a/mm/rmap.c >> +++ b/mm/rmap.c >> @@ -1564,10 +1564,11 @@ void __put_anon_vma(struct anon_vma *anon_vma) >> { >> struct anon_vma *root = anon_vma->root; >> >> - if (root != anon_vma && atomic_dec_and_test(&root->refcount)) >> + if (root != anon_vma && atomic_dec_and_test(&root->refcount)) { >> + anon_vma_free(anon_vma); >> anon_vma_free(root); >> - >> - anon_vma_free(anon_vma); >> + } else >> + anon_vma_free(anon_vma); >> } > > Why not simply move the freeing of anon_vma before the root, like: > > anon_vma_free(anon_vma); > if (root != anon_vma && atomic_dec_and_test(&root->refcount)) > anon_vma_free(root); > > ? > IMO It looks more logical to decrement root's refcounter before freeing child vma. In fact I wasn't completely sure that it is safe to do so. But after some digging, now it looks safe to me. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752128AbaFFOVu (ORCPT ); Fri, 6 Jun 2014 10:21:50 -0400 Received: from mailout2.w1.samsung.com ([210.118.77.12]:16946 "EHLO mailout2.w1.samsung.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751680AbaFFOVt (ORCPT ); Fri, 6 Jun 2014 10:21:49 -0400 X-AuditID: cbfec7f5-b7f626d000004b39-35-5391ce7ab889 Message-id: <5391CD5A.20503@samsung.com> Date: Fri, 06 Jun 2014 18:16:58 +0400 From: Andrey Ryabinin User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.5.0 MIME-version: 1.0 To: Peter Zijlstra Cc: Andrew Morton , Linus Torvalds , linux-mm@kvack.org, linux-kernel@vger.kernel.org, dvyukov@google.com, koct9i@gmail.com Subject: Re: [PATCH] mm: rmap: fix use-after-free in __put_anon_vma References: <1402054255-4930-1-git-send-email-a.ryabinin@samsung.com> <20140606115620.GS3213@twins.programming.kicks-ass.net> In-reply-to: <20140606115620.GS3213@twins.programming.kicks-ass.net> Content-type: text/plain; charset=UTF-8 Content-transfer-encoding: 7bit X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFtrPLMWRmVeSWpSXmKPExsVy+t/xa7pV5yYGG3y9z2ExZ/0aNosJD9vY LVZ2PmC1uLxrDpvFvTX/WS2O9x5gsnjU95bdgd1j56y77B4LNpV6bF6h5bHp0yR2jxMzfrN4 fN4kF8AWxWWTkpqTWZZapG+XwJUx7c5l5oLrvBUT2g+zNDD+5+pi5OSQEDCRWDXhBiuELSZx 4d56ti5GLg4hgaWMEldPTWWCcJqZJKa1/2QGqeIV0JDof3wfyObgYBFQlXj5WBUkzCagJ/Fv 1nY2kLCoQITE4wtCENWCEj8m32MBsUUENCVWtt0Hm88ssJNR4tmFHiaQhLCAs8TDj+uZQHqF BOolds1UBglzCrhIbHh/H6yEWUBdYtK8RcwQtrzE5jVvmScwCsxCsmIWkrJZSMoWMDKvYhRN LU0uKE5KzzXSK07MLS7NS9dLzs/dxAgJ9K87GJceszrEKMDBqMTDm/F+QrAQa2JZcWXuIUYJ DmYlEV625onBQrwpiZVVqUX58UWlOanFhxiZODilGhgDEpQehZ6qnv0i+2MzQ9a0KvVov1Nv J5lOUcgXO6LcsL/1EO/t9T+lpB3zt+2qe9yXfe5iwMOX65UP6vVcz+tTSdA0PzLL28SsfM6E 65tmnJ241Mrlqdv89puHZx++oVRbYaby71jZvRWZD/8flrJ/xnP/3A2d/PR+pa27Y7WUXiSJ Z24QdPigxFKckWioxVxUnAgAF8xraVICAAA= Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 06/06/14 15:56, Peter Zijlstra wrote: > On Fri, Jun 06, 2014 at 03:30:55PM +0400, Andrey Ryabinin wrote: >> While working address sanitizer for kernel I've discovered use-after-free >> bug in __put_anon_vma. >> For the last anon_vma, anon_vma->root freed before child anon_vma. >> Later in anon_vma_free(anon_vma) we are referencing to already freed anon_vma->root >> to check rwsem. >> This patch puts freeing of child anon_vma before freeing of anon_vma->root. > > Yes, I think that is right indeed. > > Very hard to hit, but valid since not all callers hold rcu_read_lock(). > >> >> Cc: stable@vger.kernel.org # v3.0+ >> Signed-off-by: Andrey Ryabinin >> --- >> mm/rmap.c | 7 ++++--- >> 1 file changed, 4 insertions(+), 3 deletions(-) >> >> diff --git a/mm/rmap.c b/mm/rmap.c >> index 9c3e773..161bffc7 100644 >> --- a/mm/rmap.c >> +++ b/mm/rmap.c >> @@ -1564,10 +1564,11 @@ void __put_anon_vma(struct anon_vma *anon_vma) >> { >> struct anon_vma *root = anon_vma->root; >> >> - if (root != anon_vma && atomic_dec_and_test(&root->refcount)) >> + if (root != anon_vma && atomic_dec_and_test(&root->refcount)) { >> + anon_vma_free(anon_vma); >> anon_vma_free(root); >> - >> - anon_vma_free(anon_vma); >> + } else >> + anon_vma_free(anon_vma); >> } > > Why not simply move the freeing of anon_vma before the root, like: > > anon_vma_free(anon_vma); > if (root != anon_vma && atomic_dec_and_test(&root->refcount)) > anon_vma_free(root); > > ? > IMO It looks more logical to decrement root's refcounter before freeing child vma. In fact I wasn't completely sure that it is safe to do so. But after some digging, now it looks safe to me.