On Tue, May 31, 2022 at 8:36 PM Julia Suvorova <jus...@redhat.com> wrote:
>
> On Tue, May 31, 2022 at 3:14 PM Ani Sinha <a...@anisinha.ca> wrote:
> >
> > On Tue, May 31, 2022 at 5:53 PM Julia Suvorova <jus...@redhat.com> wrote:
> > >
> > > On Sat, May 28, 2022 at 7:22 AM Ani Sinha <a...@anisinha.ca> wrote:
> > > >
> > > >
> > > >
> > > > On Fri, 27 May 2022, Julia Suvorova wrote:
> > > >
> > > > > The new test is run with a large number of cpus and checks if the
> > > > > core_count field in smbios_cpu_test (structure type 4) is correct.
> > > > >
> > > > > Choose q35 as it allows to run with -smp > 255.
> > > > >
> > > > > Signed-off-by: Julia Suvorova <jus...@redhat.com>
> > > > > ---
> > > > > tests/qtest/bios-tables-test.c | 35
> > > > > +++++++++++++++++++++++++++++++++-
> > > > > 1 file changed, 34 insertions(+), 1 deletion(-)
> > > > >
> > > > > diff --git a/tests/qtest/bios-tables-test.c
> > > > > b/tests/qtest/bios-tables-test.c
> > > > > index 0ba9d749a5..f2464adaa0 100644
> > > > > --- a/tests/qtest/bios-tables-test.c
> > > > > +++ b/tests/qtest/bios-tables-test.c
> > > > > @@ -100,6 +100,8 @@ typedef struct {
> > > > > smbios_entry_point smbios_ep_table;
> > > > > uint16_t smbios_cpu_max_speed;
> > > > > uint16_t smbios_cpu_curr_speed;
> > > > > + uint8_t smbios_core_count;
> > > > > + uint16_t smbios_core_count2;
> > > > > uint8_t *required_struct_types;
> > > > > int required_struct_types_len;
> > > > > QTestState *qts;
> > > > > @@ -640,8 +642,9 @@ static inline bool smbios_single_instance(uint8_t
> > > > > type)
> > > > >
> > > > > static bool smbios_cpu_test(test_data *data, uint32_t addr)
> > > > > {
> > > > > + uint8_t real_cc, expect_cc = data->smbios_core_count;
> > > > > + uint16_t real, real_cc2, expect_cc2 = data->smbios_core_count2;
> > > > > uint16_t expect_speed[2];
> > > > > - uint16_t real;
> > > >
> > > > while you are at it, I suggest renaming this to real_speed or some such
> > > > so
> > > > that its better redeable.
> > >
> > > Ok
> > >
> > > > > int offset[2];
> > > > > int i;
> > > > >
> > > > > @@ -660,6 +663,20 @@ static bool smbios_cpu_test(test_data *data,
> > > > > uint32_t addr)
> > > > > }
> > > > > }
> > > > >
> > > > > + real_cc = qtest_readb(data->qts, addr + offsetof(struct
> > > > > smbios_type_4, core_count));
> > > > > + real_cc2 = qtest_readw(data->qts, addr + offsetof(struct
> > > > > smbios_type_4, core_count2));
> > > > > +
> > > > > + if (expect_cc && (real_cc != expect_cc)) {
> > > >
> > > > I think better to say if ((expect_cc < 256) && (real_cc != expect_cc))
> > >
> > > The check is not whether it fits into the field, but whether the field
> > > is initialized.
> >
> > yes so the real_cc will contain the actual value of core count only
> > when the core count value is less than 256. This value should be the
> > same as the expect_cc (the cc value we pass). Is this not what is
> > being tested?
>
> The real_cc should always be equal to expect_cc (which is 0xFF with
> -smp 275). So if the core count is less than 256, this checks for the
> actual core counter, and if it's over, it checks if real_cc is equal
> to 0xFF, which eliminates several unnecessary comparisons. If we
> didn't initialize expect_cc in the test, the value is undefined, and
> we shouldn't check anything.
I have convinced myself that the logic is correct. You are right in
that real_cc should always be equal to expect_cc when expect_cc is set
(either to actual core count or to 0xff when number of cores > 255).
>
> Best regards, Julia Suvorova.
>
> > >
> > > > > + fprintf(stderr, "Unexpected SMBIOS CPU count: real %u expect
> > > > > %u\n",
> > > > > + real_cc, expect_cc);
> > > > > + return false;
> > > > > + }
> > > > > + if ((expect_cc == 0xFF) && (real_cc2 != expect_cc2)) {
> > > > > + fprintf(stderr, "Unexpected SMBIOS CPU count2: real %u
> > > > > expect %u\n",
> > > > > + real_cc2, expect_cc2);
> > > > > + return false;
> > > > > + }
> > > > > +
> > > > > return true;
> > > > > }
> > > > >
> > > > > @@ -905,6 +922,21 @@ static void test_acpi_q35_tcg(void)
> > > > > free_test_data(&data);
> > > > > }
> > > > >
> > > > > +static void test_acpi_q35_tcg_core_count2(void)
> > > > > +{
> > > > > + test_data data = {
> > > > > + .machine = MACHINE_Q35,
> > > > > + .variant = ".core-count2",
> > > > > + .required_struct_types = base_required_struct_types,
> > > > > + .required_struct_types_len =
> > > > > ARRAY_SIZE(base_required_struct_types),
> > > > > + .smbios_core_count = 0xFF,
> > > > > + .smbios_core_count2 = 275,
> > > > > + };
> > > > > +
> > > > > + test_acpi_one("-machine smbios-entry-point-type=64 -smp 275",
> > > > > &data);
> > > > > + free_test_data(&data);
> > > > > +}
> > > > > +
> > > > > static void test_acpi_q35_tcg_bridge(void)
> > > > > {
> > > > > test_data data;
> > > > > @@ -1787,6 +1819,7 @@ int main(int argc, char *argv[])
> > > > > qtest_add_func("acpi/piix4/pci-hotplug/off",
> > > > > test_acpi_piix4_no_acpi_pci_hotplug);
> > > > > qtest_add_func("acpi/q35", test_acpi_q35_tcg);
> > > > > + qtest_add_func("acpi/q35/core-count2",
> > > > > test_acpi_q35_tcg_core_count2);
> > > >
> > > > How about checking thread count as well in the same test or in a
> > > > different test?
> > >
> > > Maybe a different test.
> > >
> > > Best regards, Julia Suvorova.
> > >
> > > > > qtest_add_func("acpi/q35/bridge", test_acpi_q35_tcg_bridge);
> > > > > qtest_add_func("acpi/q35/multif-bridge",
> > > > > test_acpi_q35_multif_bridge);
> > > > > qtest_add_func("acpi/q35/mmio64", test_acpi_q35_tcg_mmio64);
> > > > > --
> > > > > 2.35.1
> > > > >
> > > > >
> > > >
> > >
> >
>
