Re: [RFC PATCH v2 2/3] perf/x86/intel/pt: Introduce and export pt_get_curr_event()

["Liang, Kan" <kan.liang@linux.intel.com>]

On 2022-09-21 12:45 p.m., Xiaoyao Li wrote:
> KVM supports PT_MODE_HOST_GUEST mode for Intel PT where host and guest
> have separate Intel PT configurations and they work independently.
> 
> Currently, in this mode, when both host and guest enable PT, KVM manually
> clears MSR_IA32_RTIT_CTL.TRACEEN to disable host PT so that it can
> context switch the other PT MSRs. However, PT PMI can be delivered after
> this point and before the VM-entry. As a result, the re-enabling of PT
> leads to VM-entry failure of guest.
> 
> To solve the problem, introduce and export pt_get_curr_event() for KVM
> to get current pt event. 

I don't think the current pt event is created by KVM. IIUC, the patch
basically expose the event created by the other user to KVM. That
doesn't sounds correct.

I think it should be perf's responsibility to decide which events should
be disabled, and which MSRs should be switched. Because only perf can
see all the events. The users should only see the events they created.

Thanks,
Kan

> Along with perf_event_{dis, en}able_local(),
> With them, KVM can avoid PT re-enabling in PT PMI handler.
> 
> Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
> ---
>  arch/x86/events/intel/pt.c        | 8 ++++++++
>  arch/x86/include/asm/perf_event.h | 2 ++
>  2 files changed, 10 insertions(+)
> 
> diff --git a/arch/x86/events/intel/pt.c b/arch/x86/events/intel/pt.c
> index 82ef87e9a897..62bfc45c11c9 100644
> --- a/arch/x86/events/intel/pt.c
> +++ b/arch/x86/events/intel/pt.c
> @@ -1624,6 +1624,14 @@ static void pt_event_stop(struct perf_event *event, int mode)
>  	}
>  }
>  
> +struct perf_event *pt_get_curr_event(void)
> +{
> +	struct pt *pt = this_cpu_ptr(&pt_ctx);
> +
> +	return pt->handle.event;
> +}
> +EXPORT_SYMBOL_GPL(pt_get_curr_event);
> +
>  static long pt_event_snapshot_aux(struct perf_event *event,
>  				  struct perf_output_handle *handle,
>  				  unsigned long size)
> diff --git a/arch/x86/include/asm/perf_event.h b/arch/x86/include/asm/perf_event.h
> index f6fc8dd51ef4..7c3533392cf5 100644
> --- a/arch/x86/include/asm/perf_event.h
> +++ b/arch/x86/include/asm/perf_event.h
> @@ -553,11 +553,13 @@ static inline int x86_perf_get_lbr(struct x86_pmu_lbr *lbr)
>  
>  #ifdef CONFIG_CPU_SUP_INTEL
>   extern void intel_pt_handle_vmx(int on);
> + extern struct perf_event *pt_get_curr_event(void);
>  #else
>  static inline void intel_pt_handle_vmx(int on)
>  {
>  
>  }
> +struct perf_event *pt_get_curr_event(void) { return NULL; }
>  #endif
>  
>  #if defined(CONFIG_PERF_EVENTS) && defined(CONFIG_CPU_SUP_AMD)