Messages in this thread |  | | Date | Wed, 13 Nov 2013 17:39:33 -0500 (EST) | From | Vince Weaver <> | Subject | Re: perf/tracepoint yet another lockup |
| |
On Tue, 12 Nov 2013, Peter Zijlstra wrote:
> On Mon, Nov 11, 2013 at 05:28:21PM -0500, Vince Weaver wrote: > > So are all tracepoints unsafe? Or just a subset? > > No, most should be fine. But finding the special ones seems interesting > :/ > > Disabling them all for now sure seems like the safe option.
Update on this, it was a false positive. It seems like the other crashes I was seeing were also caused by tracepoint 24, the fuzzer just found them via another code path I wasn't checking.
So from what I can tell the only troublesome tracepoint I've come across is the original irq_vectors/irq_work_exit problem.
Vince
|  |