Re: [PATCH v5 4/4] hw/acpi/aml-build: Use existing CPU topology to build PPTT table

[Igor Mammedov]

On Wed, 20 Apr 2022 13:19:34 +0800
Gavin Shan <gshan@redhat.com> wrote:

> Hi Igor,
> 
> On 4/19/22 4:54 PM, Igor Mammedov wrote:
> > On Thu, 14 Apr 2022 08:33:29 +0800
> > Gavin Shan <gshan@redhat.com> wrote:  
> >> On 4/13/22 9:52 PM, Igor Mammedov wrote:  
> >>> On Sun,  3 Apr 2022 22:59:53 +0800
> >>> Gavin Shan <gshan@redhat.com> wrote:
> >>>      
> >>>> When the PPTT table is built, the CPU topology is re-calculated, but
> >>>> it's unecessary because the CPU topology has been populated in
> >>>> virt_possible_cpu_arch_ids() on arm/virt machine.
> >>>>
> >>>> This reworks build_pptt() to avoid by reusing the existing one in
> >>>> ms->possible_cpus. Currently, the only user of build_pptt() is
> >>>> arm/virt machine.
> >>>>
> >>>> Signed-off-by: Gavin Shan <gshan@redhat.com>
> >>>> ---
> >>>>    hw/acpi/aml-build.c | 100 +++++++++++++++++---------------------------
> >>>>    1 file changed, 38 insertions(+), 62 deletions(-)
> >>>>
> >>>> diff --git a/hw/acpi/aml-build.c b/hw/acpi/aml-build.c
> >>>> index 4086879ebf..4b0f9df3e3 100644
> >>>> --- a/hw/acpi/aml-build.c
> >>>> +++ b/hw/acpi/aml-build.c
> >>>> @@ -2002,86 +2002,62 @@ void build_pptt(GArray *table_data, BIOSLinker 
> >>>> *linker, MachineState *ms,
> >>>>                    const char *oem_id, const char *oem_table_id)
> >>>>    {
> >>>>        MachineClass *mc = MACHINE_GET_CLASS(ms);
> >>>> -    GQueue *list = g_queue_new();
> >>>> -    guint pptt_start = table_data->len;
> >>>> -    guint parent_offset;
> >>>> -    guint length, i;
> >>>> -    int uid = 0;
> >>>> -    int socket;
> >>>> +    CPUArchIdList *cpus = ms->possible_cpus;
> >>>> +    int64_t socket_id = -1, cluster_id = -1, core_id = -1;
> >>>> +    uint32_t socket_offset, cluster_offset, core_offset;
> >>>> +    uint32_t pptt_start = table_data->len;
> >>>> +    int n;
> >>>>        AcpiTable table = { .sig = "PPTT", .rev = 2,
> >>>>                            .oem_id = oem_id, .oem_table_id = 
> >>>> oem_table_id };
> >>>>    
> >>>>        acpi_table_begin(&table, table_data);
> >>>>    
> >>>> -    for (socket = 0; socket < ms->smp.sockets; socket++) {
> >>>> -        g_queue_push_tail(list,
> >>>> -            GUINT_TO_POINTER(table_data->len - pptt_start));
> >>>> -        build_processor_hierarchy_node(
> >>>> -            table_data,
> >>>> -            /*
> >>>> -             * Physical package - represents the boundary
> >>>> -             * of a physical package
> >>>> -             */
> >>>> -            (1 << 0),
> >>>> -            0, socket, NULL, 0);
> >>>> -    }
> >>>> +    for (n = 0; n < cpus->len; n++) {  
> >>>      
> >>>> +        if (cpus->cpus[n].props.socket_id != socket_id) {
> >>>> +            socket_id = cpus->cpus[n].props.socket_id;  
> >>>
> >>> this relies on cpus->cpus[n].props.*_id being sorted form top to down 
> >>> levels
> >>> I'd add here and for other container_id an assert() that checks for that
> >>> specific ID goes in only one direction, to be able to detect when rule is 
> >>> broken.
> >>>
> >>> otherwise on may end up with duplicate containers silently.
> >>>      
> >>
> >> Exactly. cpus->cpus[n].props.*_id is sorted as you said in 
> >> virt_possible_cpu_arch_ids().
> >> The only user of build_pptt() is arm/virt machine. So it's fine. However, 
> >> I think I
> >> may need add comments for this in v6.
> >>
> >>       /*
> >>        * This works with the assumption that cpus[n].props.*_id has been
> >>        * sorted from top to down levels in mc->possible_cpu_arch_ids().
> >>        * Otherwise, the unexpected and duplicate containers will be 
> >> created.
> >>        */
> >>
> >> The implementation in v3 looks complicated, but comprehensive. The one
> >> in this revision (v6) looks simple, but the we're losing flexibility :)  
> > 
> > 
> > comment is not enough, as it will break silently that's why I suggested
> > sprinkling asserts() here.
> >   
> 
> I don't think it breaks anything. Duplicated PPTT entries are allowed in
> linux at least. The IDs in the duplicated PPTT entries should be same.
> Otherwise, the exposed CPU topology is really broken.

Spec doesn't say anything about allowing duplicate entries so I'd rather
avoid that (if you find a such provision in spec then put a reference
in this commit message to end discussion on duplicates).


> 
> I don't think it's harmful to add the check and assert, so I will introduce
> a helper function like below in v7. Sadly that v6 was posted before I received
> your confirm. Igor, could you please the changes, to be included into v7,
> looks good to you? The complete patch is also attached :)
> 
> +static bool pptt_entry_exists(MachineState *ms, int n, bool check_socket_id,
> +                              bool check_cluster_id, bool check_core_id)
> +{
> +    CPUArchId *cpus = ms->possible_cpus->cpus;
> +    CpuInstanceProperties *t = &cpus[n].props;
> +    CpuInstanceProperties *s;
> +    bool match;
> +    int i;
> +
> +    for (i = 0; i < n; i++) {

Wouldn't it make whole thing O(n^2) in worst case?

I suggest put asserts directly into build_pptt() and considering that
it relies on ids being sorted, do something like this:
   assert(foo_id_val > previous_id)
which will ensure that id doesn't jump back unexpectedly


> +        match = true;
> +        s = &cpus[i].props;
> +
> +        if (check_socket_id && s->socket_id != t->socket_id) {
> +            match = false;
> +        }
> +
> +        if (match && check_cluster_id && s->cluster_id != t->cluster_id) {
> +            match = false;
> +        }
> +
> +        if (match && check_core_id && s->core_id != t->core_id) {
> +            match = false;
> +        }
> +
> +        if (match) {
> +            return true;
> +        }
> +    }
> +
> +    return false;
> +}
> 
> The following assert() will be applied in build_pptt():
> 
> assert(!pptt_entry_exists(ms, n, true, false, false));           /* socket  */
> assert(!pptt_entry_exists(ms, n, true, true, false));            /* cluster */
> assert(!pptt_entry_exists(ms, n, true,
>             mc->smp_props.clusters_supported, true));             /* core    
> */
> 
> Thanks,
> Gavin
>