Re: [PATCH 2/3] perf/x86/pebs: add workaround for broken OVFL status on HSW

[Stephane Eranian]

On Tue, Mar 8, 2016 at 1:13 PM, Stephane Eranian <eranian@xxxxxxxxxx> wrote:
> Hi,
>
> On Tue, Mar 8, 2016 at 1:07 PM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>> On Tue, Mar 08, 2016 at 12:59:23PM -0800, Stephane Eranian wrote:
>>> hi,
>>>
>>> On Mon, Mar 7, 2016 at 12:25 PM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>>> >
>>> > On Mon, Mar 07, 2016 at 07:27:31PM +0100, Jiri Olsa wrote:
>>> > > On Mon, Mar 07, 2016 at 01:18:40PM +0100, Peter Zijlstra wrote:
>>> > > > On Mon, Mar 07, 2016 at 11:24:13AM +0100, Peter Zijlstra wrote:
>>> > > >
>>> > > > > I suspect Andi is having something along:
>>> > > > >
>>> > > > >  lkml.kernel.org/r/1445458568-16956-1-git-send-email-andi@xxxxxxxxxxxxxx
>>> > > > >
>>> > > > > applied to his tree.
>>> > > >
>>> > > > OK, I munged a bunch of patches together, please have a hard look at the
>>> > > > end result found in:
>>> > > >
>>> > > >   git://git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git perf/core
>>> > > >
>>>
>>> I ran this kernel on Haswell. Even with Andi's fixes the problem I identified is
>>> still there, so my patch is still needed.
>>
>> Right, your patch should be included in that kernel, or did I make a
>> royal mess of things?
>>
> No, it is as expected for the OVF PMI fix.
>
>> I put Andi's late status ack on top of your patch.
>>
Ok, I ran into a problem on Broadwell with your branch with Andi's
patches. I see
a problem which had disappeared since SandyBridge:

11551.128422] ------------[ cut here ]------------
[11551.128435] WARNING: CPU: 3 PID: 12114 at
arch/x86/events/intel/core.c:1868 intel_pmu_handle_irq+0x2da/0x4b0()
[11551.128437] perfevents: irq loop stuck!
[11551.128469]  <NMI>  [<ffffffff81663975>] dump_stack+0x4d/0x63
[11551.128479]  [<ffffffff810b8657>] warn_slowpath_common+0x97/0xe0
[11551.128482]  [<ffffffff810b8756>] warn_slowpath_fmt+0x46/0x50
[11551.128486]  [<ffffffff8100b6ca>] intel_pmu_handle_irq+0x2da/0x4b0
[11551.128491]  [<ffffffff81004569>] perf_event_nmi_handler+0x39/0x60
[11551.128494]  [<ffffffff8107be61>] nmi_handle+0x61/0x110
[11551.128497]  [<ffffffff8107c684>] default_do_nmi+0x44/0x110
[11551.128500]  [<ffffffff8107c827>] do_nmi+0xd7/0x140
[11551.128504]  [<ffffffff8166e127>] end_repeat_nmi+0x1a/0x1e
[11551.128507]  [<ffffffff81009dd6>] ? native_write_msr+0x6/0x30
[11551.128510]  [<ffffffff81009dd6>] ? native_write_msr+0x6/0x30
[11551.128514]  [<ffffffff81009dd6>] ? native_write_msr+0x6/0x30
[11551.128515]  <<EOE>>  [<ffffffff8100b385>] ?
intel_pmu_enable_event+0x215/0x230
[11551.128520]  [<ffffffff81005a0d>] x86_pmu_start+0x8d/0x120
[11551.128523]  [<ffffffff810061db>] x86_pmu_enable+0x27b/0x2f0
[11551.128527]  [<ffffffff8118d63d>] perf_pmu_enable+0x1d/0x30
[11551.128530]  [<ffffffff81191bca>] ctx_resched+0x5a/0x70
[11551.128532]  [<ffffffff81191d8c>] __perf_event_enable+0x1ac/0x210
[11551.128537]  [<ffffffff81188f81>] event_function+0xa1/0x170
[11551.128540]  [<ffffffff811899b0>] ? perf_duration_warn+0x70/0x70
[11551.128543]  [<ffffffff811899f7>] remote_function+0x47/0x60
[11551.128547]  [<ffffffff8112a178>] generic_exec_single+0xa8/0xb0
[11551.128550]  [<ffffffff811899b0>] ? perf_duration_warn+0x70/0x70
[11551.128553]  [<ffffffff811899b0>] ? perf_duration_warn+0x70/0x70
[11551.128555]  [<ffffffff8112a298>] smp_call_function_single+0xa8/0x100
[11551.128559]  [<ffffffff8118aec4>] event_function_call+0x84/0x100
[11551.128561]  [<ffffffff81191be0>] ? ctx_resched+0x70/0x70
[11551.128564]  [<ffffffff81191be0>] ? ctx_resched+0x70/0x70
[11551.128566]  [<ffffffff81188ee0>] ? perf_ctx_lock+0x30/0x30
[11551.128570]  [<ffffffff8118b050>] _perf_event_enable+0x60/0x80
[11551.128572]  [<ffffffff8118fc61>] perf_ioctl+0x271/0x3e0

The infinite loop in the irq handler!

But here it seems there is a race with a perf_events ioctl() to likely
reset the period.
I am not using the perf tool here just running a self-monitoring task.


>> Also note, Ingo merged most of those patches today, all except the top
>> 3, because Andi wanted to double check something.
>