[PATCH 00/19 v4] Fix filesystem freezing deadlocks

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

[PATCH 00/19 v4] Fix filesystem freezing deadlocks

Jan Kara
  Hello,

  here is the fourth iteration of my patches to improve filesystem freezing.
Filesystem freezing is currently racy and thus we can end up with dirty data on
frozen filesystem (see changelog patch 06 for detailed race description). This
patch series aims at fixing this.

To be able to block all places where inodes get dirtied, I've moved filesystem
freeze handling in mnt_want_write() / mnt_drop_write(). This however required
some code shuffling and changes to kern_path_create() (see patches 02-05). I
think the result is OK but opinions may differ ;). The advantage of this change
also is that all filesystems get freeze protection almost for free - even ext2
can handle freezing well now.

Another potential contention point might be patch 19. In that patch we make
freeze_super() refuse to freeze the filesystem when there are open but unlinked
files which may be impractical in some cases. The main reason for this is the
problem with handling of file deletion from fput() called with mmap_sem held
(e.g. from munmap(2)), and then there's the fact that we cannot really force
such filesystem into a consistent state... But if people think that freezing
with open but unlinked files should happen, then I have some possible
solutions in mind (maybe as a separate patchset since this is large enough).

I'm not able to hit any deadlocks, lockdep warnings, or dirty data on frozen
filesystem despite beating it with fsstress and bash-shared-mapping while
freezing and unfreezing for several hours (using ext4 and xfs) so I'm
reasonably confident this could finally be the right solution.

And for people wanting to test - this patchset is based on patch series
"Push file_update_time() into .page_mkwrite" so you'll need to pull that one
in as well.

Changes since v3:
  * added third level of freezing for fs internal purposes - hooked some
    filesystems to use it (XFS, nilfs2)
  * removed racy i_size check from filemap_mkwrite()

Changes since v2:
  * completely rewritten
  * freezing is now blocked at VFS entry points
  * two stage freezing to handle both mmapped writes and other IO

The biggest changes since v1:
  * have two counters to provide safe state transitions for SB_FREEZE_WRITE
    and SB_FREEZE_TRANS states
  * use percpu counters instead of own percpu structure
  * added documentation fixes from the old fs freezing series
  * converted XFS to use SB_FREEZE_TRANS counter instead of its private
    m_active_trans counter

                                                                Honza

CC: Alex Elder <[hidden email]>
CC: Anton Altaparmakov <[hidden email]>
CC: Ben Myers <[hidden email]>
CC: Chris Mason <[hidden email]>
CC: [hidden email]
CC: "David S. Miller" <[hidden email]>
CC: [hidden email]
CC: "J. Bruce Fields" <[hidden email]>
CC: Joel Becker <[hidden email]>
CC: KONISHI Ryusuke <[hidden email]>
CC: [hidden email]
CC: [hidden email]
CC: [hidden email]
CC: [hidden email]
CC: [hidden email]
CC: Mark Fasheh <[hidden email]>
CC: Miklos Szeredi <[hidden email]>
CC: [hidden email]
CC: OGAWA Hirofumi <[hidden email]>
CC: Steven Whitehouse <[hidden email]>
CC: "Theodore Ts'o" <[hidden email]>
CC: [hidden email]

------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Linux-NTFS-Dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/linux-ntfs-dev
Reply | Threaded
Open this post in threaded view
|

[PATCH 15/19] ntfs: Convert to new freezing mechanism

Jan Kara
Move check in ntfs_file_aio_write_nolock() to ntfs_file_aio_write() and
use new freeze protection.

CC: [hidden email]
CC: Anton Altaparmakov <[hidden email]>
Signed-off-by: Jan Kara <[hidden email]>
---
 fs/ntfs/file.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/fs/ntfs/file.c b/fs/ntfs/file.c
index c587e2d..0503e65 100644
--- a/fs/ntfs/file.c
+++ b/fs/ntfs/file.c
@@ -2084,7 +2084,6 @@ static ssize_t ntfs_file_aio_write_nolock(struct kiocb *iocb,
  if (err)
  return err;
  pos = *ppos;
- vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE);
  /* We can write back this queue in page reclaim. */
  current->backing_dev_info = mapping->backing_dev_info;
  written = 0;
@@ -2117,6 +2116,7 @@ static ssize_t ntfs_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
 
  BUG_ON(iocb->ki_pos != pos);
 
+ sb_start_write(inode->i_sb);
  mutex_lock(&inode->i_mutex);
  ret = ntfs_file_aio_write_nolock(iocb, iov, nr_segs, &iocb->ki_pos);
  mutex_unlock(&inode->i_mutex);
@@ -2125,6 +2125,7 @@ static ssize_t ntfs_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
  if (err < 0)
  ret = err;
  }
+ sb_end_write(inode->i_sb);
  return ret;
 }
 
--
1.7.1


------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Linux-NTFS-Dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/linux-ntfs-dev