On Tue, Jan 29, 2013 at 11:56:01AM -0500, Vince Weaver wrote: > On Tue, 29 Jan 2013, Jiri Olsa wrote: > > > ok, there's been perf fixies wrt enable events groups, but they > > landed in 3.8-rc1.. until that point all the group members are > > specifically enabled, so that's why your perf is working and > > mine not.. > > > > Anyway, seems like there's bug in kernel code when we have group > > with sw and hw counters. > > > > Attached path makes your test case working for me. I'll give it some > > more thoughts & testing and send it out soon. If you like to test, > > please go ahead. > > I've added a test to my perf_event_test testsuite to look for this bug. > http://www.eece.maine.edu/~vweaver/projects/perf_events/validation/ > > In my (minimal) testing so far it looks like this bug does exist in 3.2 > but does not in 2.6.35. Is anyone curious enough to do a bisect to find > out when it was introduced?
I found this started at v2.6.37 where following commit that introduced the group moving was added: perf: Complete software pmu grouping commit b04243ef7006cda301819f54ee7ce0a3632489e3 Author: Peter Zijlstra <a.p.zijls...@chello.nl> Date: Fri Sep 17 11:28:48 2010 +0200 I could not get finer bisection within 2.6.37 because the kernel deadlocked each time I ran the test. But it seems this problem is there from the introducing of the group moving. CCing more folks and lkml jirka -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/