Re: [PATCH v2 04/15] tests: acpi: q35: test for x2APIC entries in SRAT

[Igor Mammedov <imammedo@redhat.com>]

On Wed, 20 Oct 2021 10:58:55 +0200
Thomas Huth <thuth@redhat.com> wrote:

> On 20/10/2021 10.53, Igor Mammedov wrote:
> > On Wed, 20 Oct 2021 04:18:07 -0400
> > "Michael S. Tsirkin" <mst@redhat.com> wrote:
> >   
> >> On Wed, Oct 20, 2021 at 10:16:07AM +0200, Igor Mammedov wrote:  
> >>> On Tue, 19 Oct 2021 07:44:38 -0400
> >>> "Michael S. Tsirkin" <mst@redhat.com> wrote:
> >>>      
> >>>> On Tue, Oct 19, 2021 at 01:36:24PM +0200, Igor Mammedov wrote:  
> >>>>> On Tue, 19 Oct 2021 06:23:40 -0400
> >>>>> "Michael S. Tsirkin" <mst@redhat.com> wrote:
> >>>>>        
> >>>>>> On Tue, Oct 19, 2021 at 11:48:25AM +0200, Igor Mammedov wrote:  
> >>>>>>> On Mon, 18 Oct 2021 17:31:33 -0400
> >>>>>>> "Michael S. Tsirkin" <mst@redhat.com> wrote:
> >>>>>>>          
> >>>>>>>> On Thu, Sep 02, 2021 at 07:35:40AM -0400, Igor Mammedov wrote:  
> >>>>>>>>> Set -smp 1,maxcpus=288 to test for ACPI code that
> >>>>>>>>> deal with CPUs with large APIC ID (>255).
> >>>>>>>>>
> >>>>>>>>> PS:
> >>>>>>>>> Test requires KVM and in-kernel irqchip support,
> >>>>>>>>> so skip test if KVM is not available.
> >>>>>>>>>
> >>>>>>>>> Signed-off-by: Igor Mammedov <imammedo@redhat.com>
> >>>>>>>>> ---
> >>>>>>>>> v3:
> >>>>>>>>>    - add dedicated test instead of abusing 'numamem' one
> >>>>>>>>>    - add 'kvm' prefix to the test name
> >>>>>>>>>        ("Michael S. Tsirkin" <mst@redhat.com>)
> >>>>>>>>> v2:
> >>>>>>>>>    - switch to qtest_has_accel() API
> >>>>>>>>>
> >>>>>>>>> CC: thuth@redhat.com
> >>>>>>>>> CC: lvivier@redhat.com
> >>>>>>>>> ---
> >>>>>>>>>   tests/qtest/bios-tables-test.c | 17 +++++++++++++++++
> >>>>>>>>>   1 file changed, 17 insertions(+)
> >>>>>>>>>
> >>>>>>>>> diff --git a/tests/qtest/bios-tables-test.c b/tests/qtest/bios-tables-test.c
> >>>>>>>>> index 51d3a4e239..1f6779da87 100644
> >>>>>>>>> --- a/tests/qtest/bios-tables-test.c
> >>>>>>>>> +++ b/tests/qtest/bios-tables-test.c
> >>>>>>>>> @@ -1033,6 +1033,19 @@ static void test_acpi_q35_tcg_numamem(void)
> >>>>>>>>>       free_test_data(&data);
> >>>>>>>>>   }
> >>>>>>>>>   
> >>>>>>>>> +static void test_acpi_q35_kvm_xapic(void)
> >>>>>>>>> +{
> >>>>>>>>> +    test_data data;
> >>>>>>>>> +
> >>>>>>>>> +    memset(&data, 0, sizeof(data));
> >>>>>>>>> +    data.machine = MACHINE_Q35;
> >>>>>>>>> +    data.variant = ".xapic";
> >>>>>>>>> +    test_acpi_one(" -object memory-backend-ram,id=ram0,size=128M"
> >>>>>>>>> +                  " -numa node -numa node,memdev=ram0"
> >>>>>>>>> +                  " -machine kernel-irqchip=on -smp 1,maxcpus=288", &data);
> >>>>>>>>> +    free_test_data(&data);
> >>>>>>>>> +}
> >>>>>>>>> +
> >>>>>>>>>   static void test_acpi_q35_tcg_nosmm(void)
> >>>>>>>>>   {
> >>>>>>>>>       test_data data;  
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> This causes an annoying message each time I run it:
> >>>>>>>>
> >>>>>>>> qemu-system-x86_64: -accel kvm: warning: Number of hotpluggable cpus requested (288) exceeds the recommended cpus supported by KVM (240)
> >>>>>>>>
> >>>>>>>> what gives?  
> >>>>>>>
> >>>>>>> it depends on kernel, see kvm_recommended_vcpus().
> >>>>>>>
> >>>>>>> We probably should bump it on upstream kernel side
> >>>>>>> (it's much more than that in RHEL8).
> >>>>>>>
> >>>>>>> Is there anything that prevents bumping upstream kernel limits?  
> >>>>>>
> >>>>>> what should we do with the annoying warning though?  
> >>>>>
> >>>>> I'd leave it alone.
> >>>>> What do you suggest?
> >>>>>         
> >>>>
> >>>> reduce the value so a typical system does not trigger it?  
> >>>
> >>> it won't work, test needs as minimum 255 vcpus to trigger X2APIC logic
> >>> and 288 value additionally tests max limits  
> >>
> >> Add a flag to disable the warning while running this specific test?  
> > 
> > if it were qtest accel, it would be trivial but
> > I'm not aware of something similar for tcg/kvm mode.
> > Do you suggest to add to QEMU a CLI option for that?  
> 
> I think you can still use qtest_enabled(), even if -accel kvm has been 
> specified on top?
I don't see how
   accel/qtest/qtest.c:    ac->allowed = &qtest_allowed;
   ...
   accel_init_machine(): *(acc->allowed) = true;

>   Thomas
>