Date:   Fri, 17 Apr 2020 16:40:04 +0800
From:   "Xu, Like" <like.xu@...el.com>
To:     Peter Zijlstra <peterz@...radead.org>,
        Like Xu <like.xu@...ux.intel.com>
Cc:     Paolo Bonzini <pbonzini@...hat.com>, kvm@...r.kernel.org,
        Andi Kleen <ak@...ux.intel.com>,
        Jim Mattson <jmattson@...gle.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Sean Christopherson <sean.j.christopherson@...el.com>,
        Joerg Roedel <joro@...tes.org>,
        Liran Alon <liran.alon@...cle.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Liang Kan <kan.liang@...ux.intel.com>,
        Wei Wang <wei.w.wang@...el.com>, linux-kernel@...r.kernel.org,
        Ingo Molnar <mingo@...nel.org>
Subject: Re: [PATCH v9 03/10] perf/x86: Add constraint to create guest LBR
 event without hw counter

Hi Peter,

On 2020/4/10 11:03, Xu, Like wrote:
> Hi Peter,
>
> First of all, thanks for your comments!
>
> On 2020/4/10 0:37, Peter Zijlstra wrote:
>>> diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
>>> index 3bb738f5a472..e919187a0751 100644
>>> --- a/arch/x86/events/core.c
>>> +++ b/arch/x86/events/core.c
>>> @@ -74,7 +74,8 @@ u64 x86_perf_event_update(struct perf_event *event)
>>>       int idx = hwc->idx;
>>>       u64 delta;
>>>   -    if (idx == INTEL_PMC_IDX_FIXED_BTS)
>>> +    if ((idx == INTEL_PMC_IDX_FIXED_BTS) ||
>>> +        (idx == INTEL_PMC_IDX_FIXED_VLBR))
>>>           return 0;
>>>         /*
>>> @@ -1102,7 +1103,8 @@ static inline void x86_assign_hw_event(struct 
>>> perf_event *event,
>>>       hwc->last_cpu = smp_processor_id();
>>>       hwc->last_tag = ++cpuc->tags[i];
>>>   -    if (hwc->idx == INTEL_PMC_IDX_FIXED_BTS) {
>>> +    if ((hwc->idx == INTEL_PMC_IDX_FIXED_BTS) ||
>>> +        (hwc->idx == INTEL_PMC_IDX_FIXED_VLBR)) {
>>>           hwc->config_base = 0;
>>>           hwc->event_base    = 0;
>>>       } else if (hwc->idx >= INTEL_PMC_IDX_FIXED) {
>>> @@ -1233,7 +1235,8 @@ int x86_perf_event_set_period(struct perf_event 
>>> *event)
>>>       s64 period = hwc->sample_period;
>>>       int ret = 0, idx = hwc->idx;
>>>   -    if (idx == INTEL_PMC_IDX_FIXED_BTS)
>>> +    if ((idx == INTEL_PMC_IDX_FIXED_BTS) ||
>>> +        (idx == INTEL_PMC_IDX_FIXED_VLBR))
>>>           return 0;
>>>         /*
>> That seems unfortunate; can that be >= INTEL_PMC_IDX_FIXED_BTS ? If so,
>> that probably wants a comment with the definitions.
>>
>> Or otherwise check for !hwc->event_base. That should be 0 for both these
>> things.
> Yes, the !hwc->event_base looks good to me.
>>
>>> diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
>>> index 3be51aa06e67..901c82032f4a 100644
>>> --- a/arch/x86/events/intel/core.c
>>> +++ b/arch/x86/events/intel/core.c
>>> @@ -2157,6 +2157,9 @@ static void intel_pmu_disable_event(struct 
>>> perf_event *event)
>>>           return;
>>>       }
>>>   +    if (unlikely(hwc->idx == INTEL_PMC_IDX_FIXED_VLBR))
>>> +        return;
>>> +
>> Please check code-gen to see if you can cut down on brancher here;
>> there's 4 cases:
>>
>>   - vlbr
>>   - bts
>>   - fixed
>>   - gp
>>
>> perhaps you can write it like so:
>>
>> (also see 
>> https://lkml.kernel.org/r/20190828090217.GN2386@hirez.programming.kicks-ass.net 
>> )
>>
>> static void intel_pmu_enable_event(struct perf_event *event)
>> {
>>     ...
>>     int idx = hwx->idx;
>>
>>     if (idx < INTEL_PMC_IDX_FIXED) {
>>         intel_set_masks(event, idx);
>>         __x86_pmu_enable_event(hwc, ARCH_PERFMON_EVENTSEL_ENABLE);
>>     } else if (idx < INTEL_PMC_IDX_FIXED_BTS) {
>>         intel_set_masks(event, idx);
>>         intel_pmu_enable_fixed(event);
>>     } else if (idx == INTEL_PMC_IDX_FIXED_BTS) {
>>         intel_pmu_enable_bts(hwc->config);
>>     }
>>
>>     /* nothing for INTEL_PMC_IDX_FIXED_VLBR */
>> }
>>
>> That should sort the branches in order of: gp,fixed,bts,vlbr
>
> Note the current order is: bts, pebs, fixed, gp.
>
> Sure, let me try to refactor it in this way.
>>
>>>       cpuc->intel_ctrl_guest_mask &= ~(1ull << hwc->idx);
>>>       cpuc->intel_ctrl_host_mask &= ~(1ull << hwc->idx);
>>>       cpuc->intel_cp_status &= ~(1ull << hwc->idx);
>>> @@ -2241,6 +2244,9 @@ static void intel_pmu_enable_event(struct 
>>> perf_event *event)
>>>           return;
>>>       }
>>>   +    if (unlikely(hwc->idx == INTEL_PMC_IDX_FIXED_VLBR))
>>> +        return;
>>> +
>>>       if (event->attr.exclude_host)
>>>           cpuc->intel_ctrl_guest_mask |= (1ull << hwc->idx);
>>>       if (event->attr.exclude_guest)
>> idem.
> idem.
>>
>>> @@ -2595,6 +2601,15 @@ intel_bts_constraints(struct perf_event *event)
>>>       return NULL;
>>>   }
>>>   +static struct event_constraint *
>>> +intel_guest_event_constraints(struct perf_event *event)
>>> +{
>>> +    if (unlikely(is_guest_lbr_event(event)))
>>> +        return &guest_lbr_constraint;
>>> +
>>> +    return NULL;
>>> +}
>> This is a mis-nomer, it isn't just any guest_event
>
> Sure,  I'll rename it to intel_guest_lbr_event_constraints()
> instead of using it as a unified interface to get all of guest event 
> constraints.
>
>>
>>> +
>>>   static int intel_alt_er(int idx, u64 config)
>>>   {
>>>       int alt_idx = idx;
>>> @@ -2785,6 +2800,10 @@ __intel_get_event_constraints(struct 
>>> cpu_hw_events *cpuc, int idx,
>>>   {
>>>       struct event_constraint *c;
>>>   +    c = intel_guest_event_constraints(event);
>>> +    if (c)
>>> +        return c;
>>> +
>>>       c = intel_bts_constraints(event);
>>>       if (c)
>>>           return c;
>>> diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h
>>> index 1025bc6eb04f..9a62264a3068 100644
>>> --- a/arch/x86/events/perf_event.h
>>> +++ b/arch/x86/events/perf_event.h
>>> @@ -969,6 +969,20 @@ static inline bool intel_pmu_has_bts(struct 
>>> perf_event *event)
>>>       return intel_pmu_has_bts_period(event, hwc->sample_period);
>>>   }
>>>   +static inline bool is_guest_event(struct perf_event *event)
>>> +{
>>> +    if (event->attr.exclude_host && is_kernel_event(event))
>>> +        return true;
>>> +    return false;
>>> +}
>> I don't like this one, what if another in-kernel users generates an
>> event with exclude_host set ?
> Thanks for the clear attitude.
>
> How about:
> - remove the is_guest_event() to avoid potential misuse;
> - move all checks into is_guest_lbr_event() and make it dedicated:
>
> static inline bool is_guest_lbr_event(struct perf_event *event)
> {
>     if (is_kernel_event(event) &&
>         event->attr.exclude_host && needs_branch_stack(event))
>         return true;
>     return false;
> }
>
> In this case, it's safe to generate an event with exclude_host set
> and also use LBR to count guest or nothing for other in-kernel users
> because the intel_guest_lbr_event_constraints() makes LBR exclusive.
>
> For this generic usage, I may rename:
> - is_guest_lbr_event() to is_lbr_no_counter_event();
> - intel_guest_lbr_event_constraints() to 
> intel_lbr_no_counter_event_constraints();
>
> Is this acceptable to you？
> If there is anything needs to be improved, please let me know.
Do you have any preference for this ?

If you have more comments for the general idea or code details, please let 
me know.
For example, you may take a look at the interface named 
intel_pmu_create_lbr_event()
in the "[PATCH v9 07/10] KVM: x86/pmu: Add LBR feature emulation via guest 
LBR event".

If not, I'll spin the next version based on your current feedback.

Thanks,
Like Xu
>
>>> @@ -989,6 +1003,7 @@ void release_ds_buffers(void);
>>>   void reserve_ds_buffers(void);
>>>     extern struct event_constraint bts_constraint;
>>> +extern struct event_constraint guest_lbr_constraint;
>>>     void intel_pmu_enable_bts(u64 config);
>>>   diff --git a/arch/x86/include/asm/perf_event.h 
>>> b/arch/x86/include/asm/perf_event.h
>>> index e018a1cf604c..674130aca75a 100644
>>> --- a/arch/x86/include/asm/perf_event.h
>>> +++ b/arch/x86/include/asm/perf_event.h
>>> @@ -181,9 +181,19 @@ struct x86_pmu_capability {
>>>   #define GLOBAL_STATUS_UNC_OVF                BIT_ULL(61)
>>>   #define GLOBAL_STATUS_ASIF                BIT_ULL(60)
>>>   #define GLOBAL_STATUS_COUNTERS_FROZEN            BIT_ULL(59)
>>> -#define GLOBAL_STATUS_LBRS_FROZEN            BIT_ULL(58)
>>> +#define GLOBAL_STATUS_LBRS_FROZEN_BIT            58
>>> +#define GLOBAL_STATUS_LBRS_FROZEN BIT_ULL(GLOBAL_STATUS_LBRS_FROZEN_BIT)
>>>   #define GLOBAL_STATUS_TRACE_TOPAPMI            BIT_ULL(55)
>>>   +/*
>>> + * We model guest LBR event tracing as another fixed-mode PMC like BTS.
>>> + *
>>> + * We choose bit 58 (LBRS_FROZEN_BIT) which is used to indicate that 
>>> the LBR
>>> + * stack is frozen on a hardware PMI request in the PERF_GLOBAL_STATUS 
>>> msr,
>>> + * and the 59th PMC counter (if any) is not supposed to use it as well.
>> Is this saying that STATUS.58 should never be set? I don't really
>> understand the language.
> My fault, and let me make it more clearly:
>
> We choose bit 58 because it's used to indicate LBR stack frozen state
> not like other overflow conditions in the PERF_GLOBAL_STATUS msr,
> and it will not be used for any actual fixed events.
>
>>
>>> + */
>>> +#define INTEL_PMC_IDX_FIXED_VLBR GLOBAL_STATUS_LBRS_FROZEN_BIT
>>> +
>>>   /*
>>>    * Adaptive PEBS v4
>>>    */
>

Powered by blists - more mailing lists