From:   Xin Yin <yinxin.x@bytedance.com>
To:     harshadshirwadkar@gmail.com, tytso@mit.edu,
        adilger.kernel@dilger.ca
Cc:     linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org,
        Xin Yin <yinxin.x@bytedance.com>
Subject: [PATCH v2 0/2] fix blocks allocate issue during fast commit replay
Date:   Mon, 10 Jan 2022 11:51:39 +0800
Message-Id: <20220110035141.1980-1-yinxin.x@bytedance.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

when test fast_commit with xfstests generic/455, one failed case is 
after fast commit replay, fsck raise ’multiply-claimed blocks‘ issue.
one inode's etb block may share with other file.

fast commit replay procedure may allocate etb blocks for inodes, but
it may allocate blocks in use. This patch set fix this issue.

---
v2: add comments for behavior change of ext4_fc_record_regions().

Xin Yin (2):
  ext4: prevent used blocks from being allocated during fast commit
    replay
  ext4: modify the logic of ext4_mb_new_blocks_simple

 fs/ext4/ext4.h        |  2 ++
 fs/ext4/extents.c     |  4 ++++
 fs/ext4/fast_commit.c | 11 ++++++++---
 fs/ext4/mballoc.c     | 26 +++++++++++++++++---------
 4 files changed, 31 insertions(+), 12 deletions(-)

-- 
2.20.1