Hi Trond, This set is basically unchanged from the last one, aside from a bit more cleanup of unneeded arguments in patch #1. I know that you basically NAKed this set earlier this week. The issue you saw was that the generic locking codepaths never set the fl_owner value for flock locks. That's true, but nfs_flock does set this for any file_lock request that comes through it, so patch #1 is safe to apply now if you see no other issue with it. I have a patch queued for v3.16 that makes the generic flock codepaths set the fl_owner, but that's just cleanup and won't really affect how this works. The main problem that I think we need to fix soon though is the one that patch #2 fixes. An unprivileged user can trigger that BUG() and if panic_on_oops is set, then that's an unprivileged DoS at least. Jeff Layton (3): nfs4: treat lock owners as opaque values nfs4: queue free_lock_state job submission to nfsiod nfs4: turn free_lock_state into a void return operation fs/nfs/nfs4_fs.h | 26 +++++++------------- fs/nfs/nfs4proc.c | 14 +++++------ fs/nfs/nfs4state.c | 69 +++++++++++++++++++++--------------------------------- 3 files changed, 42 insertions(+), 67 deletions(-) -- 1.9.0 -- 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
