Re: [PATCH] perf: drop module reference on event init failure

[Mark Rutland <mark.rutland@arm.com>]

On Thu, Jan 08, 2015 at 11:32:13AM +0000, Will Deacon wrote:
> On Wed, Jan 07, 2015 at 02:56:51PM +0000, Mark Rutland wrote:
> > When initialising an event, perf_init_event will call try_module_get to
> > ensure that the PMU's module cannot be removed for the lifetime of the
> > event, with __free_event dropping the reference when the event is
> > finally destroyed. If something fails after the event has been
> > initialised, but before the event is installed, perf_event_alloc will
> > drop the reference on the module.
> > 
> > However, if we fail to initialise an event for some reason (e.g. we ask
> > an uncore PMU to perform sampling, and it refuses to initialise the
> > event), we do not drop the refcount. If we try to open such a bogus
> > event without a precise IDR type, we will loop over each PMU in the pmus
> > list, incrementing each of their refcounts without decrementing them.
> > 
> > This patch adds a module_put when pmu->event_init(event) fails, ensuring
> > that the refcounts are balanced in failure cases. As the innards of the
> > precise and search based initialisation look very similar, this logic is
> > hoisted out into a new helper function. While the early return for the
> > failed try_module_get is removed from the search case, this is handled
> > by the remaining return when ret is not -ENOENT.
> > 
> > Signed-off-by: Mark Rutland <mark.rutland@arm.com>
> > Cc: Will Deacon <will.deacon@arm.com>
> > Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
> > Cc: Paul Mackerras <paulus@samba.org>
> > Cc: Ingo Molnar <mingo@redhat.com>
> > Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
> > ---
> >  kernel/events/core.c | 28 ++++++++++++++++------------
> >  1 file changed, 16 insertions(+), 12 deletions(-)
> 
> Looks like a straightforward fix to me -- does it need to go to stable too?

Assuming people are using PMU drivers built as modules on stable
kernels, then yes. Otherwise it's not strictly necessary as far as I can
see.

The only such driver I'm immediately aware of is for the ARM CCN, but
it's entirely possible I missed another.

I guess this fixes c464c76eec4be587 (perf: Allow building PMU drivers as
modules).

> Acked-by: Will Deacon <will.deacon@arm.com>

Cheers.

Mark.

> 
> Will
> 
> > diff --git a/kernel/events/core.c b/kernel/events/core.c
> > index 4c1ee7f..4faccf3 100644
> > --- a/kernel/events/core.c
> > +++ b/kernel/events/core.c
> > @@ -6857,6 +6857,20 @@ void perf_pmu_unregister(struct pmu *pmu)
> >  }
> >  EXPORT_SYMBOL_GPL(perf_pmu_unregister);
> >  
> > +static int perf_try_init_event(struct pmu *pmu, struct perf_event *event)
> > +{
> > +	int ret;
> > +
> > +	if (!try_module_get(pmu->module))
> > +		return -ENODEV;
> > +	event->pmu = pmu;
> > +	ret = pmu->event_init(event);
> > +	if (ret)
> > +		module_put(pmu->module);
> > +
> > +	return ret;
> > +}
> > +
> >  struct pmu *perf_init_event(struct perf_event *event)
> >  {
> >  	struct pmu *pmu = NULL;
> > @@ -6869,24 +6883,14 @@ struct pmu *perf_init_event(struct perf_event *event)
> >  	pmu = idr_find(&pmu_idr, event->attr.type);
> >  	rcu_read_unlock();
> >  	if (pmu) {
> > -		if (!try_module_get(pmu->module)) {
> > -			pmu = ERR_PTR(-ENODEV);
> > -			goto unlock;
> > -		}
> > -		event->pmu = pmu;
> > -		ret = pmu->event_init(event);
> > +		ret = perf_try_init_event(pmu, event);
> >  		if (ret)
> >  			pmu = ERR_PTR(ret);
> >  		goto unlock;
> >  	}
> >  
> >  	list_for_each_entry_rcu(pmu, &pmus, entry) {
> > -		if (!try_module_get(pmu->module)) {
> > -			pmu = ERR_PTR(-ENODEV);
> > -			goto unlock;
> > -		}
> > -		event->pmu = pmu;
> > -		ret = pmu->event_init(event);
> > +		ret = perf_try_init_event(pmu, event);
> >  		if (!ret)
> >  			goto unlock;
> >  
> > -- 
> > 1.9.1
> > 
>