Re: perf/tracepoint: another fuzzer generated lockup

[Frederic Weisbecker]

On Sat, Nov 09, 2013 at 04:59:38PM +0100, Peter Zijlstra wrote:
> On Sat, Nov 09, 2013 at 04:27:01PM +0100, Frederic Weisbecker wrote:
> > In fact, raising an irq work from an irq work should simply be prohibited. That's not a sane
> > behaviour.
> 
> Well, it is because as you raised on IRC we could be holding locks and
> trying to avoid deadlocks. This is the very reason irq_work gets used in
> printk.
> 
> And its not a recursive run()->work()->queue() either, because as you said this
> tracepoint is in arch code _after_ work_run completes.

Yeah, indeed :(

> 
> All in all an exceedingly vexing issue.

Yep, the only sane way to solve this now is to check that signal pending, unfortunately
as you said it's not acquired that we can even just check this lockless...
--
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/