[libvirt] [PATCH 18/18] tests: Add a bunch of ppc64 cases to the cpu test

[Jiri Denemark]

On Tue, Aug 04, 2015 at 11:38:09 +0200, Andrea Bolognani wrote:
> New test cases cover the cpuCompare() and cpuBaseline() implementation.
> ---
>  tests/cputest.c                                          | 10 ++++++++++
>  tests/cputestdata/ppc64-baseline-incompatible-models.xml | 14 ++++++++++++++
>  tests/cputestdata/ppc64-baseline-same-model-result.xml   |  3 +++
>  tests/cputestdata/ppc64-baseline-same-model.xml          | 14 ++++++++++++++
>  tests/cputestdata/ppc64-host-better.xml                  |  6 ++++++
>  tests/cputestdata/ppc64-host-incomp-arch.xml             |  6 ++++++
>  tests/cputestdata/ppc64-host-no-vendor.xml               |  5 +++++
>  tests/cputestdata/ppc64-host-worse.xml                   |  6 ++++++
>  8 files changed, 64 insertions(+)
>  create mode 100644 tests/cputestdata/ppc64-baseline-incompatible-models.xml
>  create mode 100644 tests/cputestdata/ppc64-baseline-same-model-result.xml
>  create mode 100644 tests/cputestdata/ppc64-baseline-same-model.xml
>  create mode 100644 tests/cputestdata/ppc64-host-better.xml
>  create mode 100644 tests/cputestdata/ppc64-host-incomp-arch.xml
>  create mode 100644 tests/cputestdata/ppc64-host-no-vendor.xml
>  create mode 100644 tests/cputestdata/ppc64-host-worse.xml
> 
> diff --git a/tests/cputest.c b/tests/cputest.c
> index 82999f8..5f17145 100644
> --- a/tests/cputest.c
> +++ b/tests/cputest.c
> @@ -563,6 +563,13 @@ mymain(void)
>      DO_TEST_COMPARE("x86", "host", "host-no-vendor", VIR_CPU_COMPARE_IDENTICAL);
>      DO_TEST_COMPARE("x86", "host-no-vendor", "host", VIR_CPU_COMPARE_INCOMPATIBLE);
>  
> +    DO_TEST_COMPARE("ppc64", "host", "host", VIR_CPU_COMPARE_IDENTICAL);
> +    DO_TEST_COMPARE("ppc64", "host", "host-better", VIR_CPU_COMPARE_INCOMPATIBLE);
> +    DO_TEST_COMPARE("ppc64", "host", "host-worse", VIR_CPU_COMPARE_INCOMPATIBLE);
> +    DO_TEST_COMPARE("ppc64", "host", "host-incomp-arch", VIR_CPU_COMPARE_INCOMPATIBLE);
> +    DO_TEST_COMPARE("ppc64", "host", "host-no-vendor", VIR_CPU_COMPARE_IDENTICAL);
> +    DO_TEST_COMPARE("ppc64", "host-no-vendor", "host", VIR_CPU_COMPARE_INCOMPATIBLE);
> +
>      /* guest to host comparison */
>      DO_TEST_COMPARE("x86", "host", "bogus-model", VIR_CPU_COMPARE_ERROR);
>      DO_TEST_COMPARE("x86", "host", "bogus-feature", VIR_CPU_COMPARE_ERROR);
> @@ -619,6 +626,9 @@ mymain(void)
>  
>      DO_TEST_BASELINE("ppc64", "incompatible-vendors", 0, -1);
>      DO_TEST_BASELINE("ppc64", "no-vendor", 0, 0);
> +    DO_TEST_BASELINE("ppc64", "incompatible-models", 0, -1);
> +    DO_TEST_BASELINE("ppc64", "same-model", 0, 0);
> +
>      /* CPU features */
>      DO_TEST_HASFEATURE("x86", "host", "vmx", YES);
>      DO_TEST_HASFEATURE("x86", "host", "lm", YES);
> diff --git a/tests/cputestdata/ppc64-baseline-incompatible-models.xml b/tests/cputestdata/ppc64-baseline-incompatible-models.xml
> new file mode 100644
> index 0000000..7e7b9a6
> --- /dev/null
> +++ b/tests/cputestdata/ppc64-baseline-incompatible-models.xml
> @@ -0,0 +1,14 @@
> +<cpuTest>
> +<cpu>
> +  <arch>ppc64</arch>
> +  <model>POWER7</model>
> +  <vendor>IBM</vendor>
> +  <topology sockets='2' cores='4' threads='1'/>
> +</cpu>
> +<cpu>
> +  <arch>ppc64</arch>
> +  <model>POWER8</model>
> +  <vendor>IBM</vendor>
> +  <topology sockets='1' cores='1' threads='1'/>
> +</cpu>
> +</cpuTest>
> diff --git a/tests/cputestdata/ppc64-baseline-same-model-result.xml b/tests/cputestdata/ppc64-baseline-same-model-result.xml
> new file mode 100644
> index 0000000..dc0c862
> --- /dev/null
> +++ b/tests/cputestdata/ppc64-baseline-same-model-result.xml
> @@ -0,0 +1,3 @@
> +<cpu mode='custom' match='exact'>
> +  <model fallback='allow'>POWER8</model>
> +</cpu>

According to our discussion sometime last week I think PPC64 driver
should never create a CPU definition with fallback='allow' since it
doesn't make any sense for PPC. In other words, this series is missing
one patch that would make sure all CPU defs created by PPC driver honor
this.

Otherwise looks good, although you will need to add more tests making
sure we stay compatible with the old versioned CPU models.

Jirka