tracing: Final fixes for the race between open event file and deleting event

Oleg,

Is all the races that we are aware of between accessing event files and deleting events covered by these patches? I think I have them all. Was there any patches that I missed, as there were a lot of threads and lots of patches sent out, but not all were considered final.

I think I got the main ones that we decided on, but I'm not 100% sure as my INBOX is overrun by too many activities. It may be a few months before I go through them all.

I haven't pushed this to my for-next branch as I'm waiting for some feedback from others in my queue. But I did push it to my ftrace/urgent branch if you want to look at what will be going there. That branch may rebase but I wanted to get these patches tested by Fengguang before sending them to next. And I try not to rebase my for-next branch.

Thanks,

-- Steve

Oleg Nesterov (2):
debugfs: debugfs_remove_recursive() must not rely on list_empty(d_subdirs)
tracing: trace_remove_event_call() should fail if call/file is in use

Steven Rostedt (Red Hat) (3):
tracing: Add comment to describe special break case in probe_remove_event_call() tracing/kprobes: Fail to unregister if probe event files are in use tracing/uprobes: Fail to unregister if probe event files are in use

----

 fs/debugfs/inode.c           |   69 ++++++++++++++----------------------------
 include/linux/ftrace_event.h |    2 +-
 kernel/trace/trace_events.c  |   41 +++++++++++++++++++++++--
 kernel/trace/trace_kprobe.c  |   21 +++++++++----
 kernel/trace/trace_uprobe.c  |   51 +++++++++++++++++++++++--------
 5 files changed, 115 insertions(+), 69 deletions(-)