Re: [PATCH 1/2] perf: Add munmap callback

[Liang, Kan]

On 10/24/2018 8:29 PM, Peter Zijlstra wrote:
> On Wed, Oct 24, 2018 at 08:11:15AM -0700, kan.liang@xxxxxxxxxxxxxxx wrote:
> > +void perf_event_munmap(void)
> > +{
> > +	struct perf_cpu_context *cpuctx;
> > +	unsigned long flags;
> > +	struct pmu *pmu;
> > +
> > +	local_irq_save(flags);
>
> It is impossible to get here with IRQs already disabled.

I don't think so. Based on my test, IRQs are still enabled here. I once 
observed dead lock with my stress test. So I have to explicitly disable 
the IRQs here.

> > +	list_for_each_entry(cpuctx, this_cpu_ptr(&sched_cb_list), sched_cb_entry) {
> > +		pmu = cpuctx->ctx.pmu;
> > +
> > +		if (!pmu->munmap)
> > +			continue;
> > +
> > +		perf_ctx_lock(cpuctx, cpuctx->task_ctx);
> > +		perf_pmu_disable(pmu);
> > +
> > +		pmu->munmap();
> > +
> > +		perf_pmu_enable(pmu);
> > +
> > +		perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
> > +	}
> > +	local_irq_restore(flags);
> > +}
> > +
> >   static void perf_event_switch(struct task_struct *task,
> >   			      struct task_struct *next_prev, bool sched_in);
> >   
> > diff --git a/mm/mmap.c b/mm/mmap.c
> > index 5f2b2b184c60..61978ad8c480 100644
> > --- a/mm/mmap.c
> > +++ b/mm/mmap.c
> > @@ -2777,6 +2777,7 @@ int do_munmap(struct mm_struct *mm, unsigned long start, size_t len,
> >   	/*
> >   	 * Remove the vma's, and unmap the actual pages
> >   	 */
> > +	perf_event_munmap();
>
> I think that if you add the munmap hook, you should do it right and at
> least do it such that we can solve the other munmap problem.

Is this patch you mentioned?
https://lkml.org/lkml/2017/1/27/452

I will take a look and find a right place for both problems.

Thanks,
Kan
> >   	detach_vmas_to_be_unmapped(mm, vma, prev, end);
> >   	unmap_region(mm, vma, prev, start, end);
> >   
> > --
> > 2.17.1