Re: [BUG] WARNING: at kernel/trace/ring_buffer.c:3420

[Steven Rostedt <rostedt@goodmis.org>]

On Tue, 2010-03-09 at 11:35 +0800, Lai Jiangshan wrote:
> Steven Rostedt wrote:
> > On Mon, 2010-03-08 at 10:03 -0500, Steven Rostedt wrote:
> > 
> >>> ringbuffer resizing and reseting will increase the ->record_disabled
> >>> and then wait until a rcu_shced grace period passes.
> >>>
> >>> Contrarily, testing ->record_disabled should be at the same
> >>> preempt disabled critical region as writing into ringbuffer, otherwise
> >>> it will leave a window break ringbuffer resizing or reseting.
> >> So the resizing and the resetting need a synchronize_sched() after the
> >> disabling of the buffers, right?
> > 
> > Looking at the code, the synchronize_sched() is already done in
> > ring_buffer_resize, and the caller (trace.c:tracing_reset() ) also
> > disables the ring buffer and calls synchronize_sched().
> > 
> > With that, what other window is still opened (after this fix)?
> > 
> 
> This window is still opened: (RCU vs IDLE vs Tracing)
> 
> synchronize_sched() does not protect preempt_disable()/enable() for
> idle process. But tracing(function_graph, function) introduce more
> preempt_disable()/enable() for idle process. It brings windows.
> 
> I bet that this bug is not come from this window.
> (I added some strict code to RCU and did stress test,
> bug was still occurred.)

With Li's test, I was able to trigger the ring_buffer warning about
resetting while committing. I was still able to trigger it with your
patch to move the ring buffer disable code. But I did not trigger it
with both the ring buffer disable code _and_ your RCU fix.

But I was able to trigger the segfault that Li is seeing. That is most
likely a separate issue. I'm currently writing special debug code to
find out why that is happening.

Thanks!

-- Steve