Re: rb tree hrtimer lockup bug (found by perf_fuzzer)

[Thomas Gleixner]

On Mon, 31 Mar 2014, Vince Weaver wrote:

> On Mon, 31 Mar 2014, Thomas Gleixner wrote:
> > 
> > Vince, can you please disable CONFIG_DEBUG_KOBJECT_RELEASE and remove
> > all the debug patches to see whether the issue goes away?
> 
> It makes the crash on boot go away, but it is still possible to trigger

Ok. So we can leave that issue to the tty folks.

> a similar crash (hrtimer oopsing due to corrupt rbtree) using the 
> perf_fuzzer.

I did not expect that it goes away.
 
> This all started because when trying to debug the perf_fuzzer crash I 
> was told to turn on the extra debugging, and then that caused the crash at 
> boot.

I know. Thanks for the patience in decoding this nastiness.

> The fuzzer related crash happens on both a core2 and a haswell machine
> and it triggers pretty quickly when fuzzing, I'll have to go back and see 
> how repeatable it is.

Can you please reenable CONFIG_DEBUG_OBJECTS and check whether they
catch something?

Thanks,

	tglx


--
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/