Return-Path: Received: from mail-yb1-f194.google.com ([209.85.219.194]:44035 "EHLO mail-yb1-f194.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726126AbeJKQAY (ORCPT ); Thu, 11 Oct 2018 12:00:24 -0400 MIME-Version: 1.0 References: <153923113649.5546.9840926895953408273.stgit@magnolia> In-Reply-To: <153923113649.5546.9840926895953408273.stgit@magnolia> From: Amir Goldstein Date: Thu, 11 Oct 2018 11:33:57 +0300 Message-ID: Subject: Re: [PATCH v3 00/25] fs: fixes for serious clone/dedupe problems To: "Darrick J. Wong" Cc: Dave Chinner , Eric Sandeen , Linux NFS Mailing List , linux-cifs@vger.kernel.org, overlayfs , linux-xfs , Linux MM , Linux Btrfs , linux-fsdevel , ocfs2-devel@oss.oracle.com Content-Type: text/plain; charset="UTF-8" Sender: linux-nfs-owner@vger.kernel.org List-ID: On Thu, Oct 11, 2018 at 7:12 AM Darrick J. Wong wrote: > > Hi all, > > Dave, Eric, and I have been chasing a stale data exposure bug in the XFS > reflink implementation, and tracked it down to reflink forgetting to do > some of the file-extending activities that must happen for regular > writes. > > We then started auditing the clone, dedupe, and copyfile code and > realized that from a file contents perspective, clonerange isn't any > different from a regular file write. Unfortunately, we also noticed > that *unlike* a regular write, clonerange skips a ton of overflow > checks, such as validating the ranges against s_maxbytes, MAX_NON_LFS, > and RLIMIT_FSIZE. We also observed that cloning into a file did not > strip security privileges (suid, capabilities) like a regular write > would. I also noticed that xfs and ocfs2 need to dump the page cache > before remapping blocks, not after. > > In fixing the range checking problems I also realized that both dedupe > and copyfile tell userspace how much of the requested operation was > acted upon. Since the range validation can shorten a clone request (or > we can ENOSPC midway through), we might as well plumb the short > operation reporting back through the VFS indirection code to userspace. > > So, here's the whole giant pile of patches[1] that fix all the problems. > This branch is against 4.19-rc7 with Dave Chinner's XFS for-next branch. > The patch "generic: test reflink side effects" recently sent to fstests > exercises the fixes in this series. Tests are in [2]. > > --D > > [1] https://git.kernel.org/pub/scm/linux/kernel/git/djwong/xfs-linux.git/log/?h=djwong-devel > [2] https://git.kernel.org/pub/scm/linux/kernel/git/djwong/xfstests-dev.git/log/?h=djwong-devel I tested your branch with overlayfs over xfs. I did not observe any failures with -g clone except for test generic/937 which also failed on xfs in my test. I though that you forgot to mention I needed to grab xfsprogs from djwong-devel for commit e84a9e93 ("xfs_io: dedupe command should only complain if we don't dedupe anything"), but even with this change the test still fails: generic/937 - output mismatch (see /old/home/amir/src/fstests/xfstests-dev/results//generic/937.out.bad) --- tests/generic/937.out 2018-10-11 08:23:00.630938364 +0300 +++ /old/home/amir/src/fstests/xfstests-dev/results//generic/937.out.bad 2018-10-11 10:54:40.448134832 +0300 @@ -4,8 +4,7 @@ 39578c21e2cb9f6049b1cf7fc7be12a6 TEST_DIR/test-937/file2 Files 1-2 do not match (intentional) (partial) dedupe the middle blocks together -deduped XXXX/XXXX bytes at offset XXXX -XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +XFS_IOC_FILE_EXTENT_SAME: Extents did not match. Compare sections One thing that *is* different with overlayfs test is that filefrag crashes on this same test: QA output created by 937 Create the original files 35ac8d7917305c385c30f3d82c30a8f6 TEST_DIR/test-937/file1 39578c21e2cb9f6049b1cf7fc7be12a6 TEST_DIR/test-937/file2 Files 1-2 do not match (intentional) (partial) dedupe the middle blocks together XFS_IOC_FILE_EXTENT_SAME: Extents did not match. ./tests/generic/937: line 59: 19242 Floating point exception(core dumped) ${FILEFRAG_PROG} -v $testdir/file1 >> $seqres.full ./tests/generic/937: line 60: 19244 Floating point exception(core dumped) ${FILEFRAG_PROG} -v $testdir/file2 >> $seqres.full It looks like an overlayfs v4.19-rc1 regression - FIGETBSZ returns zero. I never noticed this regression before, because none of the generic tests are using filefrag. Thanks, Amir.