Re introduces a DCACHE_DENTRY_UNHASHED flag to solve a race with checking a
unhashed state vs rename. Bug found while a Cray Lustre testing but looks not a
FS specific, du comment in dcache.h
> Any filesystem which supports nfsd_operations MUST have a lookup function which,
> if it finds a directory inode with a DCACHE_DISCONNECTED dentry, will d_move
> that dentry into place and return that dentry rather than the passed one,
> typically using d_splice_alias. *
So potentially any FS will affected.
v2 removed to use a sequence lock from d_unhashed check, due large number
problems while d_unhashed checked with any spinlock hold.
---
Alexey Lyashkov (2):
rehash process protected with d_seq and d_lock locks, but VFS have access to the
cleanup of d_unhashed usage.
fs/configfs/inode.c | 2 +-
fs/dcache.c | 36 ++++++++++++++++++------------------
fs/nfs/dir.c | 2 +-
include/linux/dcache.h | 6 ++++--
4 files changed, 24 insertions(+), 22 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
