On Tue, Jun 28, 2022, Yang Weijiang wrote:
> Add helpers to check whether MSR_CORE_PERF_GLOBAL_CTRL and rdpmc
> are supported in KVM. When pmu is disabled with enable_pmu=0,
> reading MSR_CORE_PERF_GLOBAL_CTRL or executing rdpmc leads to #GP,
> so skip related tests in this case to avoid test failure.
>
> Opportunistically replace some "printf" with "report_skip" to make
> the output log clean.
Ooof, these end up dominating the patch. Can you split them to a separate prep
patch? Thanks!
> Suggested-by: Sean Christopherson <seanjc@xxxxxxxxxx>
> Signed-off-by: Yang Weijiang <weijiang.yang@xxxxxxxxx>
>
> v4:
> - Use supported_fn() to make the code nicer. [Sean]
> - Replace some of the printf with report_skip to make the results clean. [Sean]
Put the versioning info below the three dashes so that it doesn't show up in the
final changelog.
> ---
<version info goes here>
> lib/x86/processor.h | 10 ++++++++++
> x86/vmx_tests.c | 40 +++++++++++++++++++++++++++-------------
> 2 files changed, 37 insertions(+), 13 deletions(-)
>
> diff --git a/lib/x86/processor.h b/lib/x86/processor.h
> index 9a0dad6..7b6ee92 100644
> --- a/lib/x86/processor.h
> +++ b/lib/x86/processor.h
> @@ -690,4 +690,14 @@ static inline bool cpuid_osxsave(void)
> return cpuid(1).c & (1 << (X86_FEATURE_OSXSAVE % 32));
> }
>
> +static inline u8 pmu_version(void)
> +{
> + return cpuid(10).a & 0xff;
> +}
> +
> +static inline bool cpu_has_perf_global_ctrl(void)
> +{
> + return pmu_version() > 1;
> +}
> +
> #endif
> diff --git a/x86/vmx_tests.c b/x86/vmx_tests.c
> index 4d581e7..3a14cb2 100644
> --- a/x86/vmx_tests.c
> +++ b/x86/vmx_tests.c
> @@ -852,6 +852,10 @@ static bool monitor_supported(void)
> return this_cpu_has(X86_FEATURE_MWAIT);
> }
>
> +static inline bool pmu_supported(void) {
Curly brace goes on a new line.
> + return !!pmu_version();
> +}
Why not put this in processor.h? And maybe call it cpu_has_pmu()?
