Re: PATCH? trace_remove_event_call() should fail if call is active

[Steven Rostedt]

On Thu, 2013-07-04 at 00:18 +0200, Oleg Nesterov wrote:
> On 07/03, Steven Rostedt wrote:
> >
> > No, I would avoid any changes to the debugfs infrastructure.
> 
> YEs, agreed.
> 
> > OK, what about the below patch, followed by an updated version of your
> > patch. I'll send that as a reply to this one.
> 
> Steven, you do understand that I can't review the changes in this area.

I have more faith in you than you do ;-)

> 
> But at first glance, _I think_ this should work. And this is much simpler,
> ->open() blocks trace_remove_event_call() (you added TRACE_EVENT_FL_REF_MASK
> check into the next patch).

Yep.

> 
> Which tree this patch is based on? I have pulled linux-trace.git#for-next
> and I do not see tracing_open_generic_file/etc in trace_events.c.

Ug! Thanks! I posted my [for-next] series but never pushed it to my git
tree. I just pushed it now. I'm glad you told me this because I was
under the assumption that the code was already in my kernel.org repo,
and I would have pushed to Linus thinking it was already in linux-next
and would have been embarrassed if something went wrong.

> 
> I do not understand what protects call->flags, I guess there is another
> lock which I do not see in my tree?

Those flags should only be set under the event_mutex lock. But I see I
didn't do that :-)   Yeah, I need to add locks for that.

See, you can review my patch and provide valuable feedback!

-- Steve


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/