Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1560588 > unrolled thread

Re: [PATCH 05/12] x86/cqm,perf/core: Cgroup support prepare

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH 05/12] x86/cqm,perf/core: Cgroup support prepare Thomas Gleixner <tglx@linutronix.de> - 2017-01-17 14:50 +0100
    Re: [PATCH 05/12] x86/cqm,perf/core: Cgroup support prepare Shivappa Vikas <vikas.shivappa@intel.com> - 2017-01-17 21:30 +0100
      Re: [PATCH 05/12] x86/cqm,perf/core: Cgroup support prepare Thomas Gleixner <tglx@linutronix.de> - 2017-01-17 22:40 +0100

#1560588 — Re: [PATCH 05/12] x86/cqm,perf/core: Cgroup support prepare

On Fri, 6 Jan 2017, Vikas Shivappa wrote:
> @@ -741,7 +741,13 @@ static int intel_cqm_event_init(struct perf_event *event)
>  	INIT_LIST_HEAD(&event->hw.cqm_group_entry);
>  	INIT_LIST_HEAD(&event->hw.cqm_groups_entry);
>  
> -	event->destroy = intel_cqm_event_destroy;

I missed this in the first round, but tripped over it when looking at one
of the follow up patches.

How is that supposed to work?

1) intel_cqm_event_destroy() is still in the code and unused which emits a
   compiler warning, but that can obviously be ignored for a good measure.

2) How would any testing of this mess actually work?

   Not all all. Nothing ever tears down an event. So you just leave
   everything hanging around probably with dangling pointers left and
   right.

So now the 'Tests: Same as before.' in the so called changelog makes sense:

   'Same as before' means: Completely untested and broken.

Thanks,

	tglx

[toc] | [next] | [standalone]

#1560991

On Tue, 17 Jan 2017, Thomas Gleixner wrote:

> On Fri, 6 Jan 2017, Vikas Shivappa wrote:
>> @@ -741,7 +741,13 @@ static int intel_cqm_event_init(struct perf_event *event)
>>  	INIT_LIST_HEAD(&event->hw.cqm_group_entry);
>>  	INIT_LIST_HEAD(&event->hw.cqm_groups_entry);
>>
>> -	event->destroy = intel_cqm_event_destroy;
>
> I missed this in the first round, but tripped over it when looking at one
> of the follow up patches.
>
> How is that supposed to work?
>
> 1) intel_cqm_event_destroy() is still in the code and unused which emits a
>   compiler warning, but that can obviously be ignored for a good measure.
>
> 2) How would any testing of this mess actually work?
>
>   Not all all. Nothing ever tears down an event. So you just leave
>   everything hanging around probably with dangling pointers left and
>   right.

The terminate is defined in next patch. Will fix this as we dont need all this 
new api and the cgroup cqm specific structures can be freed with cgroup hooks 
instead of creating this new one.

Thanks,
Vikas

>
> So now the 'Tests: Same as before.' in the so called changelog makes sense:
>
>   'Same as before' means: Completely untested and broken.
>
> Thanks,
>
> 	tglx
>
>

[toc] | [prev] | [next] | [standalone]

#1561041

On Tue, 17 Jan 2017, Shivappa Vikas wrote:
> On Tue, 17 Jan 2017, Thomas Gleixner wrote:
> 
> > On Fri, 6 Jan 2017, Vikas Shivappa wrote:
> > > @@ -741,7 +741,13 @@ static int intel_cqm_event_init(struct perf_event
> > > *event)
> > >  	INIT_LIST_HEAD(&event->hw.cqm_group_entry);
> > >  	INIT_LIST_HEAD(&event->hw.cqm_groups_entry);
> > > 
> > > -	event->destroy = intel_cqm_event_destroy;
> > 
> > I missed this in the first round, but tripped over it when looking at one
> > of the follow up patches.
> > 
> > How is that supposed to work?
> > 
> > 1) intel_cqm_event_destroy() is still in the code and unused which emits a
> >   compiler warning, but that can obviously be ignored for a good measure.
> > 
> > 2) How would any testing of this mess actually work?
> > 
> >   Not all all. Nothing ever tears down an event. So you just leave
> >   everything hanging around probably with dangling pointers left and
> >   right.
> 
> The terminate is defined in next patch.

I know and that does not make it any better. It's broken, end of story.

Thanks,

	tglx

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web