Re: [PATCH v5 3/3] perf x86: Exposing an Uncore unit to PMON for Intel Xeon server platform

[Sudarikov, Roman]

On 11.02.2020 23:14, Greg KH wrote:
> On Tue, Feb 11, 2020 at 02:59:21PM -0500, Liang, Kan wrote:
> > On 2/11/2020 1:57 PM, Greg KH wrote:
> > > On Tue, Feb 11, 2020 at 10:42:00AM -0800, Andi Kleen wrote:
> > > > On Tue, Feb 11, 2020 at 09:15:44AM -0800, Greg KH wrote:
> > > > > On Tue, Feb 11, 2020 at 07:15:49PM +0300, roman.sudarikov@xxxxxxxxxxxxxxx wrote:
> > > > > > +static ssize_t skx_iio_mapping_show(struct device *dev,
> > > > > > +				struct device_attribute *attr, char *buf)
> > > > > > +{
> > > > > > +	struct pmu *pmu = dev_get_drvdata(dev);
> > > > > > +	struct intel_uncore_pmu *uncore_pmu =
> > > > > > +		container_of(pmu, struct intel_uncore_pmu, pmu);
> > > > > > +
> > > > > > +	struct dev_ext_attribute *ea =
> > > > > > +		container_of(attr, struct dev_ext_attribute, attr);
> > > > > > +	long die = (long)ea->var;
> > > > > > +
> > > > > > +	return sprintf(buf, "0000:%02x\n", skx_iio_stack(uncore_pmu, die));
> > > > > If "0000:" is always the "prefix" of the output of this file, why have
> > > > > it at all as you always know it is there?
> >
> > I think Roman only test with BIOS configured as single-segment. So he
> > hard-code the segment# here.
> >
> > I'm not sure if Roman can do some test with multiple-segment BIOS. If not, I
> > think we should at least print a warning here.
> >
> > > > > What is ever going to cause that to change?
> > > > I think it's just to make it a complete PCI address.
> > > Is that what this really is?  If so, it's not a "complete" pci address,
> > > is it?  If it is, use the real pci address please.
> > I think we don't need a complete PCI address here. The attr is to disclose
> > the mapping information between die and PCI BUS. Segment:BUS should be good
> > enough.
> "good enough" for today, but note that you can not change the format of
> the data in the file in the future, you would have to create a new file.
> So I suggest at least try to future-proof it as much as possible if you
> _know_ this could change.
>
> Just use the full pci address, there's no reason not to, otherwise it's
> just confusing.
>
> thanks,
>
> greg k-h
Hi Greg,

Yes, the "Segment:Bus" pair is enough to distinguish between different 
Root ports.
Please see the changes below which are to address all previous comments.

Thanks,
Roman

diff --git a/arch/x86/events/intel/uncore_snbep.c 
b/arch/x86/events/intel/uncore_snbep.c
index 96fca1ac22a4..f805fbdbbe81 100644
--- a/arch/x86/events/intel/uncore_snbep.c
+++ b/arch/x86/events/intel/uncore_snbep.c
@@ -3616,15 +3616,22 @@ skx_iio_mapping_visible(struct kobject *kobj, 
struct attribute *attr, int die)
Âstatic ssize_t skx_iio_mapping_show(struct device *dev,
ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ struct device_attribute *attr, char *buf)
Â{
+ÂÂÂ struct pci_bus *bus = NULL;
ÂÂÂÂ struct pmu *pmu = dev_get_drvdata(dev);
ÂÂÂÂ struct intel_uncore_pmu *uncore_pmu =
ÂÂÂÂÂÂÂÂ container_of(pmu, struct intel_uncore_pmu, pmu);
+ÂÂÂ int pmu_idx = uncore_pmu->pmu_idx;

ÂÂÂÂ struct dev_ext_attribute *ea =
ÂÂÂÂÂÂÂÂ container_of(attr, struct dev_ext_attribute, attr);
ÂÂÂÂ long die = (long)ea->var;

-ÂÂÂ return sprintf(buf, "0000:%02x\n", skx_iio_stack(uncore_pmu, die));
+ÂÂÂ do {
+ÂÂÂÂÂÂÂ bus = pci_find_next_bus(bus);
+ÂÂÂ } while (pmu_idx--);
+
+ÂÂÂ return sprintf(buf, "%04x:%02x\n", pci_domain_nr(bus),
+ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ skx_iio_stack(uncore_pmu, die));
Â}

Âstatic int skx_msr_cpu_bus_read(int cpu, u64 *topology)
@@ -3691,10 +3698,7 @@ static int skx_iio_get_topology(struct 
intel_uncore_type *type)
ÂÂÂÂ return 0;
Â}

-static struct attribute *uncore_empry_attr;
-
Âstatic struct attribute_group skx_iio_mapping_group = {
-ÂÂÂ .attrsÂÂÂÂÂÂÂ = &uncore_empry_attr,
ÂÂÂÂ .is_visibleÂÂÂ = skx_iio_mapping_visible,
Â};

@@ -3729,7 +3733,8 @@ static int skx_iio_set_mapping(struct 
intel_uncore_type *type)
ÂÂÂÂÂÂÂÂ return -ENOMEM;
ÂÂÂÂ }
ÂÂÂÂ for (die = 0; die < uncore_max_dies(); die++) {
-ÂÂÂÂÂÂÂ sprintf(buf, "node%ld", die);
+ÂÂÂÂÂÂÂ sprintf(buf, "die%ld", die);
+ÂÂÂÂÂÂÂ sysfs_attr_init(&eas[die].attr.attr);
ÂÂÂÂÂÂÂÂ eas[die].attr.attr.name = kstrdup(buf, GFP_KERNEL);
ÂÂÂÂÂÂÂÂ if (!eas[die].attr.attr.name) {
ÂÂÂÂÂÂÂÂÂÂÂÂ ret = -ENOMEM;
@@ -3752,6 +3757,7 @@ static int skx_iio_set_mapping(struct 
intel_uncore_type *type)
ÂÂÂÂ kfree(eas);
ÂÂÂÂ kfree(attrs);
ÂÂÂÂ kfree(type->topology);
+ÂÂÂ type->attr_update = NULL;

ÂÂÂÂ return ret;
Â}