From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756214AbdKCR4Z (ORCPT ); Fri, 3 Nov 2017 13:56:25 -0400 Received: from aserp1040.oracle.com ([141.146.126.69]:36822 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752176AbdKCR4Y (ORCPT ); Fri, 3 Nov 2017 13:56:24 -0400 Subject: Re: [PATCH 4/6] hugetlbfs: implement memfd sealing To: David Herrmann Cc: =?UTF-8?Q?Marc-Andr=c3=a9_Lureau?= , linux-mm , linux-kernel , aarcange@redhat.com, Hugh Dickins , nyc@holomorphy.com References: <20171031184052.25253-1-marcandre.lureau@redhat.com> <20171031184052.25253-5-marcandre.lureau@redhat.com> From: Mike Kravetz Message-ID: <15b59408-7c4d-bbdb-7573-5789faa05e6c@oracle.com> Date: Fri, 3 Nov 2017 10:56:11 -0700 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.4.0 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit X-Source-IP: userv0021.oracle.com [156.151.31.71] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 11/03/2017 10:41 AM, David Herrmann wrote: > Hi > > On Fri, Nov 3, 2017 at 6:12 PM, Mike Kravetz wrote: >> On 11/03/2017 10:03 AM, David Herrmann wrote: >>> Hi >>> >>> On Tue, Oct 31, 2017 at 7:40 PM, Marc-André Lureau >>> wrote: >>>> Implements memfd sealing, similar to shmem: >>>> - WRITE: deny fallocate(PUNCH_HOLE). mmap() write is denied in >>>> memfd_add_seals(). write() doesn't exist for hugetlbfs. >>>> - SHRINK: added similar check as shmem_setattr() >>>> - GROW: added similar check as shmem_setattr() & shmem_fallocate() >>>> >>>> Except write() operation that doesn't exist with hugetlbfs, that >>>> should make sealing as close as it can be to shmem support. >>> >>> SEAL, SHRINK, and GROW look fine to me. >>> >>> Regarding WRITE >> >> The commit message may not be clear. However, hugetlbfs does not support >> the write system call (or aio). The only way to modify contents of a >> hugetlbfs file is via mmap or hole punch/truncate. So, we do not really >> need to worry about those special (a)io cases for hugetlbfs. > > This is not about the write(2) syscall. Please consider this scenario > about shmem: > > You create a memfd via memfd_create() and map it writable. You now > call another kernel syscall that takes as input _any mapped page > range_. You pass your mapped memfd-addresses to it. Those syscalls > tend to use get_user_pages() to pin arbitrary user-mapped pages, as > such this also affects shmem. In this case, those pages might stay > mapped even if you munmap() your memfd! > > One example of this is using AIO-read() on any other file that > supports it, passing your mapped memfd as buffer to _read into_. The > operations supported on the memfd are irrelevant here. > The selftests contain a FUSE-based test for this, since FUSE allows > user-space to GUP pages for an arbitrary amount of time. > > The original fix for this is: > > commit 05f65b5c70909ef686f865f0a85406d74d75f70f > Author: David Herrmann > Date: Fri Aug 8 14:25:36 2014 -0700 > > shm: wait for pins to be released when sealing > > Please have a look at this. Your patches use shmem_add_seals() almost > unchanged, and as such you call into shmem_wait_for_pins() on > hugetlbfs. I would really like to see an explicit ACK that this works > on hugetlbfs. Thanks for the explanation. I missed that in your first reply. I'll look into this for hugetlbfs. -- Mike Kravetz > > Thanks > David > > -- > 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 >