The internal "API" for handling leases has a lot of problems. The main one is that on success it can return a pointer to a lease that sits on the inode's i_flock list. That pointer is only guaranteed to be valid until the i_lock is dropped, which makes it a bit dangerous to use. Also, the i_lock is held over much too much of the code, which precludes any hope of ever adding proper support for leases to distributed filesystems. This patchset is a cleanup and overhaul of the internal lease API. It fixes a number of problems in that code and makes an attempt at making that API more sane to use. The only real consumer of that API is knfsd, but this should make it easier for others to do so, reduce and clarify the spinlocking involved in handling leases, and get us a step closer toward allowing lease implementations that can block. I'm targeting this work for v3.18. Review would be welcome... Jeff Layton (10): locks: close potential race in lease_get_mtime nfsd: fix potential lease memory leak in nfs4_setlease locks: generic_delete_lease doesn't need a file_lock at all locks: clean up vfs_setlease kerneldoc comments nfsd: don't keep a pointer to the lease in nfs4_file locks: plumb an "aux" pointer into the setlease routines locks: define a lm_setup handler for leases locks: move i_lock acquisition into generic_*_lease handlers locks: move freeing of leases outside of i_lock locks: update Documentation/filesystems with lease API changes Documentation/filesystems/Locking | 21 ++-- Documentation/filesystems/vfs.txt | 7 +- fs/cifs/cifsfs.c | 7 +- fs/gfs2/file.c | 3 +- fs/locks.c | 237 +++++++++++++++++++++----------------- fs/nfs/file.c | 2 +- fs/nfs/internal.h | 2 +- fs/nfsd/nfs4state.c | 24 ++-- fs/nfsd/state.h | 1 - include/linux/fs.h | 18 +-- 10 files changed, 172 insertions(+), 150 deletions(-) -- 1.9.3 -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
