[PATCHSET,0/2] xfs: minor fixes to log recovery problems

Message

Darrick J. Wong June 16, 2021, 11:55 p.m. UTC

Hi all,

This series fixes a couple of bugs that I found in log recovery.  The
first problem is that we don't reserve space for per-AG btree expansion
when we inactivate inodes during log recovery.  If a file with shared
blocks were to be unlinked, this can result in transaction failure
because inactivation assumes that it doesn't need to reserve blocks to
free files and cannot handle a refcount btree expansion.

The second problem addressed here is that if log recovery fails due to
something that doesn't directly impact the log (like ENOSPC during
transaction allocation, or ENOMEM allocating buffers) it will leave the
log running, which means that it writes an unmount record after recovery
fails.  The /next/ mount will see a clean log and start running, even
though the metadata isn't consistent.

If you're going to start using this mess, you probably ought to just
pull from my git trees, which are linked below.

This is an extraordinary way to destroy everything.  Enjoy!
Comments and questions are, as always, welcome.

--D

kernel git tree:
https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=log-recovery-fixes-5.14
---
 fs/xfs/xfs_log.c         |    3 +++
 fs/xfs/xfs_log_recover.c |    5 ++++-
 fs/xfs/xfs_mount.c       |   10 +++++++++-
 3 files changed, 16 insertions(+), 2 deletions(-)