[v1] fix blocks allocate issue during fast commit replay

[PATCH 0/2] fix blocks allocate issue during fast commit replay

Xin Yin posted 2 patches 4 years, 5 months ago

Diff against v2
Download series mbox

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(-)

Expand all Fold all

[PATCH 0/2] fix blocks allocate issue during fast commit replay

Posted by Xin Yin 4 years, 5 months ago

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.

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