Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:1 as permitted sender) client-ip=2620:137:e000::3:1;
From:   Jiachen Zhang <zhangjiachen.jaycee@bytedance.com>
To:     Chandan Babu R <chandan.babu@oracle.com>,
        "Darrick J. Wong" <djwong@kernel.org>
Cc:     Dave Chinner <dchinner@redhat.com>,
        Allison Henderson <allison.henderson@oracle.com>,
        Zhang Tianci <zhangtianci.1997@bytedance.com>,
        Brian Foster <bfoster@redhat.com>, Ben Myers <bpm@sgi.com>,
        linux-xfs@vger.kernel.org, linux-kernel@vger.kernel.org,
        xieyongji@bytedance.com, me@jcix.top,
        Jiachen Zhang <zhangjiachen.jaycee@bytedance.com>
Subject: [PATCH v2 0/2] Fixes for ENOSPC xfs_remove
Date:   Wed, 29 Nov 2023 15:58:30 +0800
Message-Id: <20231129075832.73600-1-zhangjiachen.jaycee@bytedance.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi,

Recently, our use-case ran into 2 bugs in case doing xfs_remove when the
disk space is in-pressure, which may cause xfs shutdown and kernel crash
in the xfs log recovery procedure. Here are 2 patches to fix the
problem.

The 1st patch fixes an uninitialized variable issue.

The 2nd patch ensures the blkno in the xfs_buf is updated when doing
xfs_da3_swap_lastblock().

Compared with the V1 patchset, this V2 patchset
- directly set the *logflagsp value to make the code more robust in the
  1st commit,
- check xfs's crc-feature rather than magic in the 2nd commit, and
- fixed code style and rebased onto the master branch.


Thanks,
Jiachen

Jiachen Zhang (1):
  xfs: ensure logflagsp is initialized in xfs_bmap_del_extent_real

Zhang Tianci (1):
  xfs: update dir3 leaf block metadata after swap

 fs/xfs/libxfs/xfs_bmap.c     | 26 ++++++++++++++------------
 fs/xfs/libxfs/xfs_da_btree.c | 11 ++++++++++-
 2 files changed, 24 insertions(+), 13 deletions(-)

-- 
2.20.1