RE: [PATCH v3 2/3] perf: add arm64 smmuv3 pmu driver

[Shameerali Kolothum Thodi]

> -----Original Message-----
> From: Robin Murphy [mailto:robin.murphy@xxxxxxx]
> Sent: 11 October 2018 12:26
> To: Shameerali Kolothum Thodi <shameerali.kolothum.thodi@xxxxxxxxxx>;
> lorenzo.pieralisi@xxxxxxx
> Cc: will.deacon@xxxxxxx; mark.rutland@xxxxxxx; Guohanjun (Hanjun Guo)
> <guohanjun@xxxxxxxxxx>; John Garry <john.garry@xxxxxxxxxx>;
> pabba@xxxxxxxxxxxxxx; vkilari@xxxxxxxxxxxxxx; rruigrok@xxxxxxxxxxxxxx;
> linux-acpi@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; linux-arm-
> kernel@xxxxxxxxxxxxxxxxxxx; Linuxarm <linuxarm@xxxxxxxxxx>;
> neil.m.leeder@xxxxxxxxx
> Subject: Re: [PATCH v3 2/3] perf: add arm64 smmuv3 pmu driver
> 
> Hi Shameer,
> 
> One more thing...
> 
> On 21/09/18 16:08, Shameer Kolothum wrote:
> [...]
> > +static int smmu_pmu_probe(struct platform_device *pdev)
> > +{
> > +	struct smmu_pmu *smmu_pmu;
> > +	struct resource *res_0, *res_1;
> > +	u32 cfgr, reg_size;
> > +	u64 ceid_64[2];
> > +	int irq, err;
> > +	char *name;
> > +	struct device *dev = &pdev->dev;
> > +
> > +	smmu_pmu = devm_kzalloc(dev, sizeof(*smmu_pmu), GFP_KERNEL);
> > +	if (!smmu_pmu)
> > +		return -ENOMEM;
> > +
> > +	smmu_pmu->dev = dev;
> > +
> > +	platform_set_drvdata(pdev, smmu_pmu);
> > +	smmu_pmu->pmu = (struct pmu) {
> > +		.task_ctx_nr    = perf_invalid_context,
> > +		.pmu_enable	= smmu_pmu_enable,
> > +		.pmu_disable	= smmu_pmu_disable,
> > +		.event_init	= smmu_pmu_event_init,
> > +		.add		= smmu_pmu_event_add,
> > +		.del		= smmu_pmu_event_del,
> > +		.start		= smmu_pmu_event_start,
> > +		.stop		= smmu_pmu_event_stop,
> > +		.read		= smmu_pmu_event_read,
> > +		.attr_groups	= smmu_pmu_attr_grps,
> > +	};
> > +
> > +	res_0 = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> > +	smmu_pmu->reg_base = devm_ioremap_resource(dev, res_0);
> 
> We still need to solve the resource-claiming issue when one (or both) of
> the PMCG pages belongs to the parent device's register space. I recall
> we chucked a few nascent ideas about before; did anyone manage to come
> up with anything concrete?

Right. We had an early version of an evaluation board where we had this issue, 
but this has been fixed in an updated revision and is not a priority for now. 

Agree that this is an issue as the spec doesnât forbid using parent SMMU register
space and it looks like not an easy one to solve either. The initial idea was setting
the PMCG as a child dev, but that didnât help.

I had an off list discussion with Lorenzo on this, but nothing concrete.

Lorenzo, 

Please update if you have any new ideas/thoughts on this.

Thanks,
Shameer

> Robin.