On Fri, Feb 10, 2012 at 05:07:56PM +0100, Jiri Olsa wrote:
> On Fri, Feb 10, 2012 at 09:45:20AM -0500, Steven Rostedt wrote:
> > On Fri, 2012-02-10 at 08:27 -0500, Steven Rostedt wrote:
> > > On Tue, 2012-02-07 at 20:44 +0100, Jiri Olsa wrote:
> > > > hi,
> > > > here's another version of perf support for function trace
> > > > with filter. 
> > > 
> > > OK, I've applied all of these and I'm now going to run them through my
> > > test suite before sending them to Ingo. I'll let you know if anything
> > > goes wrong.
> > 
> > OK, I build booted and installed a kernel with your patches. I also
> > installed the latest perf code from linus/master. Then I ran the
> > following:
> > 
> > # perf record -e ftrace:function ls
> > 
> > and then ran
> > 
> > # perf report
> > 
> > It went into an infinite loop of "Processing time ordered events".
> > 
> > Is this expected?
> > 
> > Although perf script works fine.
> 
> yep, I was using perf script for testing
> I get the same error for perf report, also if you run it like:
> 
> [root@dhcp-26-214 perf]# ./perf report --stdio
> Warning:
> Processed 82392 events and lost 1 chunks!
> 
> Check IO/CPU overload!
> 
> # ========
> # captured on: Fri Feb 10 19:15:37 2012
> # hostname : dhcp-26-214.brq.redhat.com
> # os release : 3.3.0-rc3+
> # perf version : 3.3.rc3.1274.g135d69
> # arch : x86_64
> # nrcpus online : 2
> # nrcpus avail : 2
> # cpudesc : Intel(R) Core(TM)2 Duo CPU T9400 @ 2.53GHz
> # cpuid : GenuineIntel,6,23,10
> # total memory : 1969896 kB
> # cmdline : /home/jolsa/tip/tools/perf/perf record -e ftrace:function ls 
> # event : name = ftrace:function, type = 2, config = 0x1, config1 = 0x0,
> # config2 = 0x0, excl_usr = 0, excl_kern = 0, id = { 3, 4 }
> # HEADER_CPU_TOPOLOGY info available, use -I to display
> # ========
> #
> Warning: Timestamp below last timeslice flush

Stephane Eranian has reported a similar problem but perhaps not with
the same cause.

I need to have a look.