Re: [PATCH v11 8/8] perf: ARM DynamIQ Shared Unit PMU support

[Suzuki K Poulose]

On 07/03/18 21:36, Saravana Kannan wrote:
> On 03/07/2018 06:59 AM, Suzuki K Poulose wrote:
> > Hi Saravana,
> >
> > Sorry for the late response, I was out on vacation.
> >
> > On 05/03/18 22:10, Saravana Kannan wrote:
> > > On 03/05/2018 02:59 AM, Mark Rutland wrote:
> > > > On Fri, Mar 02, 2018 at 11:19:56AM -0800, Saravana Kannan wrote:
> > > > > On 03/02/2018 02:42 AM, Mark Rutland wrote:
> > > > > > It's important to note that the DSU PMU's event_init() ensures events
> > > > > > are affine to a single CPU, and the perf core code serializes
> > > > > > operations
> > > > > > on those events via the context lock.
> > > > >
> > > > > Ah, I see that now. Thanks!
> > > > >
> > > > > > Therefore, two CPUs *won't* try to access the registers
> > > > > > simultaneously.
> > > > >
> > > > > Right, and this driver seems to be going through a lot of work to
> > > > > make sure
> > > > > all events are read in one CPU.
> > > > >
> > > > > Do you even have an upstream target where there are multiple DSU's in a
> > > > > system?
> > > >
> > > > I have no idea, though I do beleive that it's possible for a system to
> > > > have multiple DSUs.
> > > >
> > > > > If not, we can simplify a ton of this code (no hotplug notifiers, no
> > > > > migrating PMUs, no SMP calls, etc) by just adding a spinlock and
> > > > > letting any
> > > > > CPU read these DSU counters.
> > > >
> > > > Regardless of whether we allow arbitrary CPUs to read the counters,
> > > > other logic still needs to be CPU affine, and we'll still need hotplug
> > > > notifiers and event migration.
> > >
> > > If you have to support multiple DSUs in a system, then the need is
> > > obvious. But if you don't have to support multiple DSU, it's not
> > > obvious to me on why you still need CPU affining or hotplug notifiers.
> > > Could you please provide me pointers for general understanding?
> >
> > We need to support multiple DSUs as such configurations are possible.
> >
> > > > I am not necessarily opposed to allowing read() calls from associated
> > > > CPUs, but as before, I'll leave that to Suzuki.
> >
> > I am fine with reading the registers from any of the associated CPUs.
>
> If that's the case, can you please use my patch?

The DSU driver is already mainline. Please feel free to send a patch with me in Cc.

> And if it looks good to you, give an Ack and ask Peter to pull it in as you'd be the user?

As mentioned above, when you send a patch, please do keep me in Cc, and will do the
needful.

Thanks
Suzuki