On Tue, Apr 17, 2012 at 07:01:29PM +0100, Al Viro wrote:
> It isn't. Hell knows - I wonder if taking s_vfs_rename_mutex in all cases
> in lock_rename() would be the right thing to do; it would remove the
> problem, but might cost us too much contention...
Actually, it's even worse. ext4_move_extents() locks a _pair_ of
->i_mutex (having checked that both are non-directories first). In
i_ino order. So the only plausible ordering would be
* directories by tree order (with s_vfs_rename_mutex held to
stabilize the tree topology)
* non-directories after all directories, ordered in some consistent
way. Which would have to be by inumber if we want to leave ext4 code
as-is.
Bruce: for now I'm dropping that patch. We _might_ take ext4
mutex_inode_double_lock() into fs/namei.c and have it used by
vfs_rename_other(), but I'm not convinced that this is the right
thing to do. Is there any other sane way to deal with nfsd problem?
i_mutex is already used for more things than I'd like...
Linus: I've removed that sucker from the queue; it should propagate to
git.kernel.org in a few. Could you pull the rest? Same place, the stats
are below:
Shortlog:
Al Viro (12):
nfsd: fix b0rken error value for setattr on read-only mount
nfsd: fix error values returned by nfsd4_lockt() when nfsd_open() fails
nfsd: fix endianness breakage in TEST_STATEID handling
nfsd: fix error value on allocation failure in nfsd4_decode_test_stateid()
nfsd: fix compose_entry_fh() failure exits
ext4: fix endianness breakage in ext4_split_extent_at()
btrfs: btrfs_root_readonly() broken on big-endian
ocfs2: ->l_next_free_req breakage on big-endian
ocfs: ->rl_used breakage on big-endian
ocfs2: ->rl_count endianness breakage
ocfs2: ->e_leaf_clusters endianness breakage
lockd: fix the endianness bug
Diffstat:
fs/btrfs/ctree.h | 2 +-
fs/ext4/extents.c | 2 +-
fs/lockd/clnt4xdr.c | 2 +-
fs/lockd/clntxdr.c | 2 +-
fs/nfsd/nfs3xdr.c | 22 ++++++++--------------
fs/nfsd/nfs4proc.c | 7 ++++---
fs/nfsd/nfs4state.c | 23 +++++++++--------------
fs/nfsd/nfs4xdr.c | 4 ++--
fs/ocfs2/alloc.c | 2 +-
fs/ocfs2/refcounttree.c | 12 ++++++------
fs/ocfs2/suballoc.c | 4 ++--
11 files changed, 36 insertions(+), 46 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
