From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-13.5 required=3.0 tests=DKIMWL_WL_MED,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS,MAILING_LIST_MULTI, MENTIONS_GIT_HOSTING,SPF_PASS,URIBL_BLOCKED,USER_IN_DEF_DKIM_WL autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 4BF2EC43387 for ; Fri, 14 Dec 2018 16:59:15 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id F0743206BA for ; Fri, 14 Dec 2018 16:59:15 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="gvvibs+u" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729997AbeLNQ7N (ORCPT ); Fri, 14 Dec 2018 11:59:13 -0500 Received: from mail-it1-f193.google.com ([209.85.166.193]:51808 "EHLO mail-it1-f193.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729409AbeLNQ7N (ORCPT ); Fri, 14 Dec 2018 11:59:13 -0500 Received: by mail-it1-f193.google.com with SMTP id x19so9822640itl.1 for ; Fri, 14 Dec 2018 08:59:12 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc; bh=pX4aR1lu7wJBeZZTeK3sSpuD9jJX/CI6ffE/yAvqj3E=; b=gvvibs+uTlUhw6zoMluwumFRecjSVusX13oex2s1l30bZGtESKpC7LGObtj6WyUzpj 8Z9pnTmEaIwpa6kZrnqpCXiLsa2lbgACDqeOffk3eK9cOtkN5hSO+0ha0Npx8W/WX8xD bLGsppEfeRPpp6EV1SpVqnP2q3/hWJrLgKXzzb+tu5HS8XiB0ob/c99F3Q3iqcLNdZtV OIRb3I5pWHttJjaQshzGOICrU5+RN4lq2l1YyMO33+3NALXWzxyUwH54I39zH8KgN+OX ZDt1wbACx/tlYghIlYfHspWPWiGuqzD9LasIEMMfAOcgKxTiTJ4554NBwiiyjqbMDxmG aPfw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=pX4aR1lu7wJBeZZTeK3sSpuD9jJX/CI6ffE/yAvqj3E=; b=hW+NFntPbmvUudNOrKo4MBaf5MxiumCVY0KByLZh3r1hhdHiqivTn2+Uw3HmMnj53T rGJcpv0P4hRjemzXPoxKYHcoTjkTFo27WNyO1oIgZY6/tJgNSoS9E46CMG4L8RRp7dNE X6SvNTHoM7Jof6l/+X+wd8RGGe5vLv37eZeEQ5hiSqHpoSUZ8KwbmT4dwURnuxCLkT4C BENgrcHt3WXswDRBcApTck4l3CXgogELw+Ik1DXR20j2yWiKEOOA806yJ96hIN4K+lPd 59USxA6FyNHTfefrXSCnCzBJwE1L1eC5Q3+1nqO3Gr/T0bspy3XVYk9vjfnifPx4An7R zhtA== X-Gm-Message-State: AA+aEWbiy3e62YUC6S3bYXNJTwB5tAWITkzGfKKUCSEnOurcAQIgmNrk sV5T+kEK6jBNAabzVvuctiV2834yqON1pmO5nmlErQ== X-Google-Smtp-Source: AFSGD/UG8x7sSNhNd7UzdfLJhXQpMe+VO0FVtGApvkejcciXZSsIZektQj3DJbyJXfi+YDUuYxtAWy6EoAYvzb79lbk= X-Received: by 2002:a24:b20e:: with SMTP id u14mr3885372ite.12.1544806751640; Fri, 14 Dec 2018 08:59:11 -0800 (PST) MIME-Version: 1.0 References: <0000000000004ea80b057cfae21e@google.com> <20181214132836.GE5343@dhcp22.suse.cz> <20181214135419.GG5343@dhcp22.suse.cz> <20181214165258.GI5343@dhcp22.suse.cz> In-Reply-To: <20181214165258.GI5343@dhcp22.suse.cz> From: Dmitry Vyukov Date: Fri, 14 Dec 2018 17:59:00 +0100 Message-ID: Subject: Re: general protection fault in watchdog To: Michal Hocko Cc: Pablo Neira Ayuso , Jozsef Kadlecsik , Florian Westphal , David Miller , NetFilter , netdev , syzbot , Andrew Morton , LKML , Paul McKenney , Tetsuo Handa , rafael.j.wysocki@intel.com, syzkaller-bugs , vkuznets@redhat.com, Linux-MM Content-Type: text/plain; charset="UTF-8" Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Dec 14, 2018 at 5:53 PM Michal Hocko wrote: > > On Fri 14-12-18 15:31:44, Dmitry Vyukov wrote: > > On Fri, Dec 14, 2018 at 2:54 PM Michal Hocko wrote: > > > > > > On Fri 14-12-18 14:42:33, Dmitry Vyukov wrote: > > > > On Fri, Dec 14, 2018 at 2:28 PM Michal Hocko wrote: > > > > > > > > > > On Fri 14-12-18 14:11:05, Dmitry Vyukov wrote: > > > > > > On Fri, Dec 14, 2018 at 1:51 PM syzbot > > > > > > wrote: > > > > > > > > > > > > > > Hello, > > > > > > > > > > > > > > syzbot found the following crash on: > > > > > > > > > > > > > > HEAD commit: f5d582777bcb Merge branch 'for-linus' of git://git.kernel... > > > > > > > git tree: upstream > > > > > > > console output: https://syzkaller.appspot.com/x/log.txt?x=16aca143400000 > > > > > > > kernel config: https://syzkaller.appspot.com/x/.config?x=c8970c89a0efbb23 > > > > > > > dashboard link: https://syzkaller.appspot.com/bug?extid=7713f3aa67be76b1552c > > > > > > > compiler: gcc (GCC) 8.0.1 20180413 (experimental) > > > > > > > syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1131381b400000 > > > > > > > C reproducer: https://syzkaller.appspot.com/x/repro.c?x=13bae593400000 > > > > > > > > > > > > > > IMPORTANT: if you fix the bug, please add the following tag to the commit: > > > > > > > Reported-by: syzbot+7713f3aa67be76b1552c@syzkaller.appspotmail.com > > > > > > > > > > > > +linux-mm for memcg question > > > > > > > > > > > > What the repro does is effectively just > > > > > > setsockopt(EBT_SO_SET_ENTRIES). This eats all machine memory and > > > > > > causes OOMs. Somehow it also caused the GPF in watchdog when it > > > > > > iterates over task list, perhaps some scheduler code leaves a dangling > > > > > > pointer on OOM failures. > > > > > > > > > > > > But what bothers me is a different thing. syzkaller test processes are > > > > > > sandboxed with a restrictive memcg which should prevent them from > > > > > > eating all memory. do_replace_finish calls vmalloc, which uses > > > > > > GFP_KERNEL, which does not include GFP_ACCOUNT (GFP_KERNEL_ACCOUNT > > > > > > does). And page alloc seems to change memory against memcg iff > > > > > > GFP_ACCOUNT is provided. > > > > > > Am I missing something or vmalloc is indeed not accounted (DoS)? I see > > > > > > some explicit uses of GFP_KERNEL_ACCOUNT, e.g. the one below, but they > > > > > > seem to be very sparse. > > > > > > > > > > > > static void *seq_buf_alloc(unsigned long size) > > > > > > { > > > > > > return kvmalloc(size, GFP_KERNEL_ACCOUNT); > > > > > > } > > > > > > > > > > > > Now looking at the code I also don't see how kmalloc(GFP_KERNEL) is > > > > > > accounted... Which makes me think I am still missing something. > > > > > > > > > > You are not missing anything. We do not account all allocations and you > > > > > have to explicitly opt-in by __GFP_ACCOUNT. This is a deliberate > > > > > decision. If the allocation is directly controlable by an untrusted user > > > > > and the memory is associated with a process life time then this looks > > > > > like a good usecase for __GFP_ACCOUNT. If an allocation outlives a > > > > > process then there the flag should be considered with a great care > > > > > because oom killer is not able to resolve the memcg pressure and so the > > > > > limit enforcement is not effective. > > > > > > > > Interesting. > > > > I understand that namespaces, memcg's and processes (maybe even > > > > threads) can have arbitrary overlapping. But I naively thought that in > > > > canonical hierarchical cases it should all somehow work. > > > > Question 1: is there some other, stricter sandboxing mechanism? > > > > > > I do not think so > > > > > > > We try > > > > to sandbox syzkaller processes with everything available , because > > > > these OOMs usually leads either to dead machines or hang/stall false > > > > positives, which are nasty. > > > > > > Which is a useful test on its own. If you are able to trigger the global > > > OOM from a restricted environment then you have a good candidate to > > > consider a new __GFP_ACCOUNT user. > > > > > > > Question 2: this is a simple DoS vector, right? If I put a container > > > > into a 1MB memcg, it can still eat arbitrary amount of non-pagable > > > > kernel memory? > > > > > > As I've said. If there is a direct vector to allocated an unbounded > > > amount of memory from the userspace (trusted users aside) then yes this > > > sounds like a DoS to me. > > > > +netfilter maintainers for this easy DoS vector > > short story: vmalloc in do_replace_finish allows unbounded memory > > allocation not accounted to memcg > > Haven't we discussed that recently? I just gave an executive summary to the new people I added to CC. > > Looks pretty unbounded: > > > > [ 763.451796] syz-executor681: vmalloc: allocation failure, allocated > > 1027440640 of 1879052288 bytes, mode:0x6000c0(GFP_KERNEL), > > nodemask=(null) > > > > But how does this play with what you said about memory outliving > > process? Netfilter tables can definitely outlive the process, they are > > attached to net ns. > > Not very well, but it is not hopeless either. The charged memory > wouldn't go away with oom victims so it will stay there until somebody > destroys those objects. On the other hand any process within that memcg > would be killed so a DoS should be somehow contained. > > > Also, am I reading this correctly that potentially thousands of kernel > > memory allocations need to be converted to ACCOUNT? I mean for small > > ones we maybe care less, but they _should_ be accounted. They can also > > build up, or just simply allow small repeated allocations. > > > > GFP_KERNEL Referenced in 11070 files: > > https://elixir.bootlin.com/linux/latest/ident/GFP_KERNEL > > > > GFP_KERNEL_ACCOUNT Referenced in 19 files: > > https://elixir.bootlin.com/linux/latest/ident/GFP_KERNEL_ACCOUNT > > Yes, we do not care about all kernel objects. Most of them are contained > by some high-level objects already. The purpose of the kmem accounting > is to limit those objects that can grow really large and are reclaimable > in some way. I see, thanks for the explanations.