Re: [PATCH v2 1/4] hw/acpi/aml-build: Only generate cluster node in PPTT when specified

[Yicong Yang via <qemu-devel@nongnu.org>]

Hi Yanan,

On 2022/10/31 14:56, wangyanan (Y) wrote:
> Hi Yicong,
> 
> On 2022/10/27 11:26, Yicong Yang wrote:
>> From: Yicong Yang <yangyicong@hisilicon.com>
>>
>> Currently we'll always generate a cluster node no matter user has
>> specified '-smp clusters=X' or not. Cluster is an optional level
>> and will participant the building of Linux scheduling domains and
>> only appears on a few platforms. It's unncessary to always build
>> it which cannot reflect the real topology on platforms have no
>> cluster and to avoid affecting the linux scheduling domains in the
>> VM. So only generate it when user specified explicitly.
>>
>> Tested qemu-system-aarch64 with `-smp 8` and linux 6.1-rc1, without
>> this patch:
>> estuary:/sys/devices/system/cpu/cpu0/topology$ cat cluster_*
>> ff    # cluster_cpus
>> 0-7    # cluster_cpus_list
>> 56    # cluster_id
>>
>> with this patch:
>> estuary:/sys/devices/system/cpu/cpu0/topology$ cat cluster_*
>> ff    # cluster_cpus
>> 0-7    # cluster_cpus_list
>> 36    # cluster_id, with no cluster node kernel will make it to
>>       physical package id
>>
>> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
>> ---
>>   hw/acpi/aml-build.c   | 2 +-
>>   hw/core/machine-smp.c | 3 +++
>>   include/hw/boards.h   | 2 ++
>>   qemu-options.hx       | 2 ++
>>   4 files changed, 8 insertions(+), 1 deletion(-)
>>
>> diff --git a/hw/acpi/aml-build.c b/hw/acpi/aml-build.c
>> index e6bfac95c7..aab73af66d 100644
>> --- a/hw/acpi/aml-build.c
>> +++ b/hw/acpi/aml-build.c
>> @@ -2030,7 +2030,7 @@ void build_pptt(GArray *table_data, BIOSLinker *linker, MachineState *ms,
>>                   0, socket_id, NULL, 0);
>>           }
>>   -        if (mc->smp_props.clusters_supported) {
>> +        if (mc->smp_props.clusters_supported && ms->smp.build_cluster) {
>>               if (cpus->cpus[n].props.cluster_id != cluster_id) {
>>                   assert(cpus->cpus[n].props.cluster_id > cluster_id);
>>                   cluster_id = cpus->cpus[n].props.cluster_id;
>> diff --git a/hw/core/machine-smp.c b/hw/core/machine-smp.c
>> index b39ed21e65..5d37e8d07a 100644
>> --- a/hw/core/machine-smp.c
>> +++ b/hw/core/machine-smp.c
>> @@ -158,6 +158,9 @@ void machine_parse_smp_config(MachineState *ms,
>>       ms->smp.threads = threads;
>>       ms->smp.max_cpus = maxcpus;
>>   +    if (config->has_clusters)
>> +        ms->smp.build_cluster = true;
>> +
>>       /* sanity-check of the computed topology */
>>       if (sockets * dies * clusters * cores * threads != maxcpus) {
>>           g_autofree char *topo_msg = cpu_hierarchy_to_string(ms);
>> diff --git a/include/hw/boards.h b/include/hw/boards.h
>> index 311ed17e18..c53f047b90 100644
>> --- a/include/hw/boards.h
>> +++ b/include/hw/boards.h
>> @@ -305,6 +305,7 @@ typedef struct DeviceMemoryState {
>>    * @cores: the number of cores in one cluster
>>    * @threads: the number of threads in one core
>>    * @max_cpus: the maximum number of logical processors on the machine
>> + * @build_cluster: build cluster topology or not
>>    */
>>   typedef struct CpuTopology {
>>       unsigned int cpus;
>> @@ -314,6 +315,7 @@ typedef struct CpuTopology {
>>       unsigned int cores;
>>       unsigned int threads;
>>       unsigned int max_cpus;
>> +    bool build_cluster;
> build_cluster seems a variable defined specifically for ACPI PPTT
> generation. It may not be proper to place it in the generic struct
> CpuTopology which only holds topo members.
> 
> What about a more generic variable in struct SMPCompatProps
> together with @clusters_supported. Something like below:
> 

ok. will follow the suggestion. Thanks for the snippet :)

> diff --git a/include/hw/boards.h b/include/hw/boards.h
> index 1f57ee8ca2..8db0706d5d 100644
> --- a/include/hw/boards.h
> +++ b/include/hw/boards.h
> @@ -130,11 +130,14 @@ typedef struct {
>   * @prefer_sockets - whether sockets are preferred over cores in smp parsing
>   * @dies_supported - whether dies are supported by the machine
>   * @clusters_supported - whether clusters are supported by the machine
> + * @has_clusters - whether clusters is explicitly specified in the user
> + *    provided SMP configuration.
>   */
>  typedef struct {
>      bool prefer_sockets;
>      bool dies_supported;
>      bool clusters_supported;
> +    bool has_clusters;
>  } SMPCompatProps;
> 
>>   } CpuTopology;
>>     /**
>> diff --git a/qemu-options.hx b/qemu-options.hx
>> index eb38e5dc40..0a710e7be3 100644
>> --- a/qemu-options.hx
>> +++ b/qemu-options.hx
>> @@ -342,6 +342,8 @@ SRST
>>       were preferred over threads), however, this behaviour is considered
>>       liable to change. Prior to 6.2 the preference was sockets over cores
>>       over threads. Since 6.2 the preference is cores over sockets over threads.
>> +    The cluster topology will only be generated if explicitly specified
>> +    by the "-cluster" option.
> no "-cluster" option, only "-smp" :)

ok.

>>       For example, the following option defines a machine board with 2 sockets
>>       of 1 core before 6.2 and 1 socket of 2 cores after 6.2:
> Maybe better to add a note at *end* of the doc about -smp like:
> 
> Note: The cluster topology will only be generated in ACPI and exposed
> to guest if it's explicitly specified in -smp.
> 

will do.

Thanks.