Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 30 Sep 2022 15:06:47 -0300
From:   Jason Gunthorpe <jgg@nvidia.com>
To:     Dan Williams <dan.j.williams@intel.com>
Cc:     Jan Kara <jack@suse.cz>, Dave Chinner <david@fromorbit.com>,
        akpm@linux-foundation.org, Matthew Wilcox <willy@infradead.org>,
        "Darrick J. Wong" <djwong@kernel.org>,
        Christoph Hellwig <hch@lst.de>,
        John Hubbard <jhubbard@nvidia.com>,
        linux-fsdevel@vger.kernel.org, nvdimm@lists.linux.dev,
        linux-xfs@vger.kernel.org, linux-mm@kvack.org,
        linux-ext4@vger.kernel.org
Subject: Re: [PATCH v2 05/18] xfs: Add xfs_break_layouts() to the inode
 eviction path
Message-ID: <YzcwN67+QOqXpvfg@nvidia.com>
References: <YyuQI08LManypG6u@nvidia.com>
 <20220923001846.GX3600936@dread.disaster.area>
 <632d00a491d0d_4a67429488@dwillia2-xfh.jf.intel.com.notmuch>
 <20220923021012.GZ3600936@dread.disaster.area>
 <20220923093803.nroajmvn7twuptez@quack3>
 <20220925235407.GA3600936@dread.disaster.area>
 <20220926141055.sdlm3hkfepa7azf2@quack3>
 <63362b4781294_795a6294f0@dwillia2-xfh.jf.intel.com.notmuch>
 <20220930134144.pd67rbgahzcb62mf@quack3>
 <63372dcbc7f13_739029490@dwillia2-mobl3.amr.corp.intel.com.notmuch>
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <63372dcbc7f13_739029490@dwillia2-mobl3.amr.corp.intel.com.notmuch>
MIME-Version: 1.0
X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1
X-MS-Exchange-AntiSpam-MessageData-0: =?us-ascii?Q?vwSCPACWCehZ61pX859K6ldd/hYj/z2HSvFfD3HhEaRqRqg5Chq2tAc3OF94?=
 =?us-ascii?Q?4rDevhDZ82ydw7OEbiaFszyylFmTZgdxhy0cowMlpBal/wrZTEAm+OLMxtQN?=
 =?us-ascii?Q?Sg7BfK9KPNFgNBnDTxAJkaPmiEwqar7LdFi1w/LXQ9BkIw3Gpg9XFMtMjKLv?=
 =?us-ascii?Q?r/TNxmTJ2F/rFLbzR7FgB/bfRry49FwkZJe6mxUVdqza3piIdOobomUAzC5k?=
 =?us-ascii?Q?Pt+8kqrW/iJfrpypPTszBuylqwo29wrndNWR0l43asDgzOZSqSlmUMWaBpFb?=
 =?us-ascii?Q?ZkdH1eCuA3z7oFlL89yU08u5sDu8IPkNDaSYPmcV5FUsRv2JdaCT9hYcIkTM?=
 =?us-ascii?Q?VRasf/uKhRUt6D4HFdOl5vvNx5iHmTNx/lFWZ7BHSqtM2s72Log0928m/O1J?=
 =?us-ascii?Q?aqdIrYfh9HZRVwIlZNtJziqLt5fAc4Rt9ke5JV0hFihqzueOzpetM8wrPMS3?=
 =?us-ascii?Q?6hDH1N+YoWuElwK82KjYZT0D/WoLbZt2LP6kYG34+corJp7zhilq1jGFDU+v?=
 =?us-ascii?Q?A4BF5K3CJ1QGHXuakF25jeStid5v9IWtQcXGEwuKuDMSf5aMpLrYHHyG2rRO?=
 =?us-ascii?Q?T6p7rbc4oQ/EJNijJoLCtXsKi4WekoPh0GrGiZHTN+rffCX9hhUbmdZFQwn5?=
 =?us-ascii?Q?KtOnyvvp/q97E4KHc2/U051xcA4C0ib80/2a7Lwi/4BzptLrZKetJAxo6bbx?=
 =?us-ascii?Q?SLjPYu+Bs9yqXbatxBwv9oLlRFGfUoLQNKlCp9Wum8/0gusYe6jfYL7ZDuOq?=
 =?us-ascii?Q?co/dLjVfnzJHuRmMtAFl866pstH+mi92rWvHyZyDAtrBXvCcdJn6ejofI/DQ?=
 =?us-ascii?Q?u6cOM4/A3ecBRM4LAtp2/BIa159LcQXvGeZaV7rHwvhZY49mxxI8SMEoGpCm?=
 =?us-ascii?Q?tMtTvziRX0TLHfIzJW6jM/ANsbHIkgxATZ/9fmJDYN6VPQ6oYXOSv+Irjat0?=
 =?us-ascii?Q?9Hhk6wQwNCMMZtnUEhCe8LIulnbUSIHpMqgC/NWyp9xFz0/a+B1Eb4PvZNwD?=
 =?us-ascii?Q?q12cvp/QapxxbjJHm7MExHv6rrF1qf8J2ZINaJcJobo/GX9sHKQoMYMWByLJ?=
 =?us-ascii?Q?Nx/OR2RiUiAotVAAb9dZU9iI7nqeK7DIx9NyogzfOBpZh9tl52qilGfmK+54?=
 =?us-ascii?Q?nrWBrnmSenYh9OJ5st4enANMBB77mMD7KzafVqaSH/mQdI7MhEhqVEnGv8RY?=
 =?us-ascii?Q?lXDTfnC2CFVW5qXe4Plb9jl91oWpaQXxURYgHJpsP/vvp4JOyYO6az2cRLlf?=
 =?us-ascii?Q?wizX3PyIGouXwyHwlecMhKvhCTZXvHJoDJiwExC9U54fZg6UYrLQl2tkTQCF?=
 =?us-ascii?Q?A/5Au4NUGDqgcCcnKX/+d6cpZ6QAHALtH3bOz2mzaHNVDlrc2yez/vr1ktN6?=
 =?us-ascii?Q?fUfKlWm6Qot0jvX6CDoa1sLkgRuG83a33gMKHfXcQ3FYC+TPYMmzu87+515B?=
 =?us-ascii?Q?uE7svM5Fggl4HxcUGbLdGTtF5d2ga5gvHpyB7KGv/rEbebQhaLDdBGeSJ/T+?=
 =?us-ascii?Q?vRuRZPzYUMzTFxxI/sAVACaI1bLDb6m1UFjQUYDw8l30JL56brxyqklUY6WF?=
 =?us-ascii?Q?KmTDyD9/a8x5ejdaRE0=3D?=
X-OriginatorOrg: Nvidia.com
X-MS-Exchange-CrossTenant-Network-Message-Id: 4c5a7adb-a01c-4197-8f3c-08daa30e8b2e
X-MS-Exchange-CrossTenant-AuthSource: MN2PR12MB4192.namprd12.prod.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 30 Sep 2022 18:06:48.7876
 (UTC)
X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted
X-MS-Exchange-CrossTenant-Id: 43083d15-7273-40c1-b7db-39efd9ccc17a
X-MS-Exchange-CrossTenant-MailboxType: HOSTED
X-MS-Exchange-CrossTenant-UserPrincipalName: FfY8d+o+4W779KJcZ1jNXczR7gBEpBhAnShus+6HbNQbh7Tk5OksH2geP+UJoo1J
X-MS-Exchange-Transport-CrossTenantHeadersStamped: PH7PR12MB6809
X-Spam-Status: No, score=-1.2 required=5.0 tests=BAYES_00,DKIMWL_WL_HIGH,
        DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,FORGED_SPF_HELO,
        RCVD_IN_DNSWL_NONE,RCVD_IN_MSPIKE_H2,SPF_HELO_PASS,SPF_NONE
        autolearn=no autolearn_force=no version=3.4.6
X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on
        lindbergh.monkeyblade.net
Precedence: bulk
List-ID: <linux-ext4.vger.kernel.org>
X-Mailing-List: linux-ext4@vger.kernel.org

On Fri, Sep 30, 2022 at 10:56:27AM -0700, Dan Williams wrote:
> Jan Kara wrote:
> [..]
> > I agree this is doable but there's the nasty sideeffect that inode reclaim
> > may block for abitrary time waiting for page pinning. If the application
> > that has pinned the page requires __GFP_FS memory allocation to get to a
> > point where it releases the page, we even have a deadlock possibility.
> > So it's better than the UAF issue but still not ideal.
> 
> I expect VMA pinning would have similar deadlock exposure if pinning a
> VMA keeps the inode allocated. Anything that puts a page-pin release
> dependency in the inode freeing path can potentially deadlock a reclaim
> event that depends on that inode being freed.

I think the desire would be to go from the VMA to an inode_get and
hold the inode reference for the from the pin_user_pages() to the
unpin_user_page(), ie prevent it from being freed in the first place.

It is a fine idea, the trouble is just the high complexity to get
there.

However, I wonder if the trucate/hole punch paths have the same
deadlock problem?

I agree with you though, given the limited options we should convert
the UAF into an unlikely deadlock.

Jason