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=-5.3 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,MAILING_LIST_MULTI,NICE_REPLY_A,SPF_HELO_NONE, SPF_PASS,USER_AGENT_SANE_1 autolearn=no 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 3D3BAC47082 for ; Wed, 26 May 2021 15:39:25 +0000 (UTC) Received: from us-smtp-delivery-124.mimecast.com (unknown [216.205.24.124]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id ABA14613D2 for ; Wed, 26 May 2021 15:39:24 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org ABA14613D2 Authentication-Results: mail.kernel.org; dmarc=fail (p=quarantine dis=none) header.from=samba.org Authentication-Results: mail.kernel.org; spf=tempfail smtp.mailfrom=linux-audit-bounces@redhat.com Received: from mimecast-mx01.redhat.com (mimecast-mx01.redhat.com [209.132.183.4]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-361-buXzrAJ3Mbyf_pZT_MkElQ-1; Wed, 26 May 2021 11:39:18 -0400 X-MC-Unique: buXzrAJ3Mbyf_pZT_MkElQ-1 Received: from smtp.corp.redhat.com (int-mx05.intmail.prod.int.phx2.redhat.com [10.5.11.15]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx01.redhat.com (Postfix) with ESMTPS id 0FE7A802690; Wed, 26 May 2021 15:39:14 +0000 (UTC) Received: from colo-mx.corp.redhat.com (colo-mx02.intmail.prod.int.phx2.redhat.com [10.5.11.21]) by smtp.corp.redhat.com (Postfix) with ESMTPS id B227861F5E; Wed, 26 May 2021 15:39:13 +0000 (UTC) Received: from lists01.pubmisc.prod.ext.phx2.redhat.com (lists01.pubmisc.prod.ext.phx2.redhat.com [10.5.19.33]) by colo-mx.corp.redhat.com (Postfix) with ESMTP id A194555345; Wed, 26 May 2021 15:39:12 +0000 (UTC) Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.rdu2.redhat.com [10.11.54.6]) by lists01.pubmisc.prod.ext.phx2.redhat.com (8.13.8/8.13.8) with ESMTP id 14QFHpIM000310 for ; Wed, 26 May 2021 11:17:52 -0400 Received: by smtp.corp.redhat.com (Postfix) id D257B21CAC6C; Wed, 26 May 2021 15:17:51 +0000 (UTC) Received: from mimecast-mx02.redhat.com (mimecast01.extmail.prod.ext.rdu2.redhat.com [10.11.55.17]) by smtp.corp.redhat.com (Postfix) with ESMTPS id CDD2C21CAC6B for ; Wed, 26 May 2021 15:17:51 +0000 (UTC) Received: from us-smtp-1.mimecast.com (us-smtp-delivery-1.mimecast.com [207.211.31.120]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id B58BE8556F0 for ; Wed, 26 May 2021 15:17:51 +0000 (UTC) Received: from hr2.samba.org (hr2.samba.org [144.76.82.148]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-299-edNzYj_zMJexeRatSFFwaw-1; Wed, 26 May 2021 11:17:49 -0400 X-MC-Unique: edNzYj_zMJexeRatSFFwaw-1 Received: from [127.0.0.2] (localhost [127.0.0.1]) by hr2.samba.org with esmtpsa (TLS1.3:ECDHE_RSA_CHACHA20_POLY1305:256) (Exim) id 1llvI3-0000gU-9J; Wed, 26 May 2021 15:17:47 +0000 To: Paul Moore , Pavel Begunkov References: <162163367115.8379.8459012634106035341.stgit@sifl> <162163379461.8379.9691291608621179559.stgit@sifl> <162219f9-7844-0c78-388f-9b5c06557d06@gmail.com> <8943629d-3c69-3529-ca79-d7f8e2c60c16@kernel.dk> <0a668302-b170-31ce-1651-ddf45f63d02a@gmail.com> From: Stefan Metzmacher Subject: Re: [RFC PATCH 2/9] audit,io_uring,io-wq: add some basic audit support to io_uring Message-ID: <18823c99-7d65-0e6f-d508-a487f1b4b9e7@samba.org> Date: Wed, 26 May 2021 17:17:46 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.8.1 MIME-Version: 1.0 In-Reply-To: X-Mimecast-Impersonation-Protect: Policy=CLT - Impersonation Protection Definition; Similar Internal Domain=false; Similar Monitored External Domain=false; Custom External Domain=false; Mimecast External Domain=false; Newly Observed Domain=false; Internal User Name=false; Custom Display Name List=false; Reply-to Address Mismatch=false; Targeted Threat Dictionary=false; Mimecast Threat Dictionary=false; Custom Threat Dictionary=false X-Scanned-By: MIMEDefang 2.78 on 10.11.54.6 X-loop: linux-audit@redhat.com X-Mailman-Approved-At: Wed, 26 May 2021 11:32:31 -0400 Cc: Jens Axboe , selinux@vger.kernel.org, linux-security-module@vger.kernel.org, linux-audit@redhat.com, Kumar Kartikeya Dwivedi , linux-fsdevel@vger.kernel.org, io-uring@vger.kernel.org, Alexander Viro X-BeenThere: linux-audit@redhat.com X-Mailman-Version: 2.1.12 Precedence: junk List-Id: Linux Audit Discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: linux-audit-bounces@redhat.com Errors-To: linux-audit-bounces@redhat.com X-Scanned-By: MIMEDefang 2.79 on 10.5.11.15 Authentication-Results: relay.mimecast.com; auth=pass smtp.auth=CUSA124A263 smtp.mailfrom=linux-audit-bounces@redhat.com X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Language: en-US Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Am 26.05.21 um 16:38 schrieb Paul Moore: > On Wed, May 26, 2021 at 6:19 AM Pavel Begunkov wrote: >> On 5/26/21 3:04 AM, Paul Moore wrote: >>> On Tue, May 25, 2021 at 9:11 PM Jens Axboe wrote: >>>> On 5/24/21 1:59 PM, Paul Moore wrote: >>>>> That said, audit is not for everyone, and we have build time and >>>>> runtime options to help make life easier. Beyond simply disabling >>>>> audit at compile time a number of Linux distributions effectively >>>>> shortcut audit at runtime by adding a "never" rule to the audit >>>>> filter, for example: >>>>> >>>>> % auditctl -a task,never >>>> >>>> As has been brought up, the issue we're facing is that distros have >>>> CONFIG_AUDIT=y and hence the above is the best real world case outside >>>> of people doing custom kernels. My question would then be how much >>>> overhead the above will add, considering it's an entry/exit call per op. >>>> If auditctl is turned off, what is the expectation in turns of overhead? >>> >>> I commented on that case in my last email to Pavel, but I'll try to go >>> over it again in a little more detail. >>> >>> As we discussed earlier in this thread, we can skip the req->opcode >>> check before both the _entry and _exit calls, so we are left with just >>> the bare audit calls in the io_uring code. As the _entry and _exit >>> functions are small, I've copied them and their supporting functions >>> below and I'll try to explain what would happen in CONFIG_AUDIT=y, >>> "task,never" case. >>> >>> + static inline struct audit_context *audit_context(void) >>> + { >>> + return current->audit_context; >>> + } >>> >>> + static inline bool audit_dummy_context(void) >>> + { >>> + void *p = audit_context(); >>> + return !p || *(int *)p; >>> + } >>> >>> + static inline void audit_uring_entry(u8 op) >>> + { >>> + if (unlikely(audit_enabled && audit_context())) >>> + __audit_uring_entry(op); >>> + } >> >> I'd rather agree that it's my cycle-picking. The case I care about >> is CONFIG_AUDIT=y (because everybody enable it), and io_uring >> tracing _not_ enabled at runtime. If enabled let them suffer >> the overhead, it will probably dip down the performance >> >> So, for the case I care about it's two of >> >> if (unlikely(audit_enabled && current->audit_context)) >> >> in the hot path. load-test-jump + current, so it will >> be around 7x2 instructions. We can throw away audit_enabled >> as you say systemd already enables it, that will give >> 4x2 instructions including 2 conditional jumps. > > We've basically got it down to the equivalent of two > "current->audit_context != NULL" checks in the case where audit is > built into the kernel but disabled at runtime, e.g. CONFIG_AUDIT=y and > "task,never". I'm at a loss for how we can lower the overhead any > further, but I'm open to suggestions. > >> That's not great at all. And that's why I brought up >> the question about need of pre and post hooks and whether >> can be combined. Would be just 4 instructions and that is >> ok (ish). > > As discussed previously in this thread that isn't really an option > from an audit perspective. > >>> We would need to check with the current security requirements (there >>> are distro people on the linux-audit list that keep track of that >>> stuff), but looking at the opcodes right now my gut feeling is that >>> most of the opcodes would be considered "security relevant" so >>> selective auditing might not be that useful in practice. It would >>> definitely clutter the code and increase the chances that new opcodes >>> would not be properly audited when they are merged. >> >> I'm curious, why it's enabled by many distros by default? Are there >> use cases they use? > > We've already talked about certain users and environments where audit > is an important requirement, e.g. public sector, health care, > financial institutions, etc.; without audit Linux wouldn't be an > option for these users, at least not without heavy modification, > out-of-tree/ISV patches, etc. I currently don't have any direct ties > to any distros, "Enterprise" or otherwise, but in the past it has been > my experience that distros much prefer to have a single kernel build > to address the needs of all their users. In the few cases I have seen > where a second kernel build is supported it is usually for hardware > enablement. I'm sure there are other cases too, I just haven't seen > them personally; the big distros definitely seem to have a strong > desire to limit the number of supported kernel configs/builds. > >> Tempting to add AUDIT_IOURING=default N, but won't work I guess > > One of the nice things about audit is that it can give you a history > of what a user did on a system, which is very important for a number > of use cases. If we selectively disable audit for certain subsystems > we create a blind spot in the audit log, and in the case of io_uring > this can be a very serious blind spot. I fear that if we can't come > to some agreement here we will need to make io_uring and audit > mutually exclusive at build time which would be awful; forcing many > distros to either make a hard choice or carry out-of-tree patches. I'm wondering why it's not enough to have the native auditing just to happen. E.g. all (I have checked RECVMSG,SENDMSG,SEND and CONNECT) socket related io_uring opcodes already go via security_socket_{recvmsg,sendmsg,connect}() IORING_OP_OPENAT* goes via do_filp_open() which is in common with the open[at[2]]() syscalls and should also trigger audit_inode() and security_file_open(). So why is there anything special needed for io_uring (now that the native worker threads are used)? Is there really any io_uring opcode that bypasses the security checks the corresponding native syscall would do? If so, I think that should just be fixed... Additional LSM based restrictions could be hooked into the io_check_restriction() path and setup at io_uring_setup() or early io_uring_register() time. What do you think? metze -- Linux-audit mailing list Linux-audit@redhat.com https://listman.redhat.com/mailman/listinfo/linux-audit