Messages in this thread | | | Date | Fri, 10 Feb 2012 07:10:46 +0100 | From | Ingo Molnar <> | Subject | Re: Hmm.. WARN_ON_ONCE(cpuc->enabled) |
| |
* David Ahern <dsahern@gmail.com> wrote:
> On 02/09/2012 04:46 PM, Linus Torvalds wrote: > > Ok, so I get the appended on my Core i5 when I did a "perf record -f > > -e cycles:pp make -j" on the kernel build. > > > > What's up, guys? I haven't tried that particular perf run in a while, > > so it might have been going on for quite some time. > > Perhaps already fixed by > https://lkml.org/lkml/2012/2/4/42 > > It's not in your tree; but in tip/urgent.
That patch only fixed the symptom - there was an underlying bug that is fixed in perf/urgent as well, via:
f39d47ff819e: perf: Fix double start/stop in x86_pmu_start()
I committed it two days ago - so it narrowly missed -rc3. Note that no actual misbehavior is known to be caused by this bug, beyond the annoying WARN_ONCE() messages and a potential slight mis-measurement.
I'll send those fixes to Linus in a couple of hours.
Thanks,
Ingo
| |