[PATCH v3] topology: make core_mask include at least cluster_siblings

[Dietmar Eggemann]

On 09/03/2022 19:26, Darren Hart wrote:
> On Wed, Mar 09, 2022 at 01:50:07PM +0100, Dietmar Eggemann wrote:
>> On 08/03/2022 18:49, Darren Hart wrote:
>>> On Tue, Mar 08, 2022 at 05:03:07PM +0100, Dietmar Eggemann wrote:
>>>> On 08/03/2022 12:04, Vincent Guittot wrote:
>>>>> On Tue, 8 Mar 2022 at 11:30, Will Deacon <will at kernel.org> wrote:
>>
>> [...]
>>
>>>> IMHO, if core_mask weight is 1, MC will be removed/degenerated anyway.
>>>>
>>>> This is what I get on my Ampere Altra (I guess I don't have the ACPI
>>>> changes which would let to a CLS sched domain):
>>>>
>>>> # cat /sys/kernel/debug/sched/domains/cpu0/domain*/name
>>>> DIE
>>>> NUMA
>>>> root at oss-altra01:~# zcat /proc/config.gz | grep SCHED_CLUSTER
>>>> CONFIG_SCHED_CLUSTER=y
>>>
>>> I'd like to follow up on this. Would you share your dmidecode BIOS
>>> Information section?
>>
>> # dmidecode -t 0
>> # dmidecode 3.2
>> Getting SMBIOS data from sysfs.
>> SMBIOS 3.2.0 present.
>>
>> Handle 0x0000, DMI type 0, 26 bytes
>> BIOS Information
>> 	Vendor: Ampere(TM)
>> 	Version: 0.9.20200724
>> 	Release Date: 2020/07/24
>> 	ROM Size: 7680 kB
>> 	Characteristics:
>> 		PCI is supported
>> 		BIOS is upgradeable
>> 		Boot from CD is supported
>> 		Selectable boot is supported
>> 		ACPI is supported
>> 		UEFI is supported
>> 	BIOS Revision: 5.15
>> 	Firmware Revision: 0.6
>>
> 
> Thank you, I'm following internally and will get with you.

Looks like in my PPTT, the `Processor Hierarchy Nodes` which represents
cluster nodes have no valid `ACPI Processor ID`.

Example for CPU0:

cpu_node-:

[1B9Ch 7068   1]           Subtable Type : 00 [Processor Hierarchy Node]
[1B9Dh 7069   1]                       Length : 1C
[1B9Eh 7070   2]                     Reserved : 0000
[1BA0h 7072   4]        Flags (decoded below) : 0000001A
                            Physical package : 0
                     ACPI Processor ID valid : 1           <-- valid !!!
                       Processor is a thread : 0
                              Node is a leaf : 1
                    Identical Implementation : 1
[1BA4h 7076   4]                       Parent : 00001B88  <-- parent !!!
[1BA8h 7080   4]            ACPI Processor ID : 00001200 [1BACh 7084
4]      Private Resource Number : 00000002
[1BB0h 7088   4]             Private Resource : 00001B58
[1BB4h 7092   4]             Private Resource : 00001B70

cluster_node (cpu_node->parent):

[1B88h 7048   1]           Subtable Type : 00 [Processor Hierarchy Node]
[1B89h 7049   1]                       Length : 14
[1B8Ah 7050   2]                     Reserved : 0000
[1B8Ch 7052   4]        Flags (decoded below) : 00000010
                            Physical package : 0
                     ACPI Processor ID valid : 0       <-- not valid !!!
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                       Processor is a thread : 0
                              Node is a leaf : 0
                    Identical Implementation : 1
[1B90h 7056   4]                       Parent : 000001C8
[1B94h 7060   4]            ACPI Processor ID : 00000000
[1B98h 7064   4]      Private Resource Number : 00000000

The code which checks this is:

int find_acpi_cpu_topology_cluster(unsigned int cpu)
{
    ....
    if (cluster_node->flags & ACPI_PPTT_ACPI_PROCESSOR_ID_VALID)
        retval = cluster_node->acpi_processor_id;
    else
       retval = ACPI_PTR_DIFF(cluster_node, table);

The else patch just returns distinct values for each CPU, so there is no
sub-grouping of CPUs which can lead to a CLS SD.