kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Sean Christopherson <seanjc@google.com>
To: Like Xu <like.xu.linux@gmail.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>, kvm@vger.kernel.org
Subject: Re: [kvm-unit-tests PATCH v3 10/13] x86/pmu: Update testcases to cover Intel Arch PMU Version 1
Date: Wed, 5 Oct 2022 22:35:34 +0000	[thread overview]
Message-ID: <Yz4GtqyPIMCMsUEl@google.com> (raw)
In-Reply-To: <20220819110939.78013-11-likexu@tencent.com>

On Fri, Aug 19, 2022, Like Xu wrote:
> From: Like Xu <likexu@tencent.com>
> 
> For most unit tests, the basic framework and use cases which test
> any PMU counter do not require any changes, except for two things:
> 
> - No access to registers introduced only in PMU version 2 and above;
> - Expanded tolerance for testing counter overflows
>   due to the loss of uniform control of the gloabl_ctrl register
> 
> Adding some pmu_version() return value checks can seamlessly support
> Intel Arch PMU Version 1, while opening the door for AMD PMUs tests.

Phrase this as a command so that it's crystal clear that this is what the patch
does, as opposed to what the patch _can_ do.

> Signed-off-by: Like Xu <likexu@tencent.com>
> ---
>  x86/pmu.c | 64 +++++++++++++++++++++++++++++++++++++------------------
>  1 file changed, 43 insertions(+), 21 deletions(-)
> 
> diff --git a/x86/pmu.c b/x86/pmu.c
> index 25fafbe..826472c 100644
> --- a/x86/pmu.c
> +++ b/x86/pmu.c
> @@ -125,14 +125,19 @@ static struct pmu_event* get_counter_event(pmu_counter_t *cnt)
>  
>  static void global_enable(pmu_counter_t *cnt)
>  {
> -	cnt->idx = event_to_global_idx(cnt);
> +	if (pmu_version() < 2)

Helper please.

> +		return;
>  
> +	cnt->idx = event_to_global_idx(cnt);
>  	wrmsr(MSR_CORE_PERF_GLOBAL_CTRL, rdmsr(MSR_CORE_PERF_GLOBAL_CTRL) |
>  			(1ull << cnt->idx));
>  }
>  
>  static void global_disable(pmu_counter_t *cnt)
>  {
> +	if (pmu_version() < 2)
> +		return;
> +
>  	wrmsr(MSR_CORE_PERF_GLOBAL_CTRL, rdmsr(MSR_CORE_PERF_GLOBAL_CTRL) &
>  			~(1ull << cnt->idx));
>  }
> @@ -301,7 +306,10 @@ static void check_counter_overflow(void)
>  	count = cnt.count;
>  
>  	/* clear status before test */
> -	wrmsr(MSR_CORE_PERF_GLOBAL_OVF_CTRL, rdmsr(MSR_CORE_PERF_GLOBAL_STATUS));
> +	if (pmu_version() > 1) {

Should be a helper to use from an earlier patch.

Hmm, looking forward, maybe have an upper level helper?  E.g.

  void pmu_clear_global_status_safe(void)
  {
	if (!exists)
		return

	wrmsr(...);
  }

Ignore this suggestion if these checks go away in the future.

> +		wrmsr(MSR_CORE_PERF_GLOBAL_OVF_CTRL,
> +		      rdmsr(MSR_CORE_PERF_GLOBAL_STATUS));
> +	}
>  
>  	report_prefix_push("overflow");
>  
> @@ -327,13 +335,21 @@ static void check_counter_overflow(void)
>  			cnt.config &= ~EVNTSEL_INT;
>  		idx = event_to_global_idx(&cnt);
>  		__measure(&cnt, cnt.count);
> -		report(cnt.count == 1, "cntr-%d", i);
> +
> +		report(check_irq() == (i % 2), "irq-%d", i);
> +		if (pmu_version() > 1)

Helper.

> +			report(cnt.count == 1, "cntr-%d", i);
> +		else
> +			report(cnt.count < 4, "cntr-%d", i);
> +
> +		if (pmu_version() < 2)

Helper.

> +			continue;
> +
>  		status = rdmsr(MSR_CORE_PERF_GLOBAL_STATUS);
>  		report(status & (1ull << idx), "status-%d", i);
>  		wrmsr(MSR_CORE_PERF_GLOBAL_OVF_CTRL, status);
>  		status = rdmsr(MSR_CORE_PERF_GLOBAL_STATUS);
>  		report(!(status & (1ull << idx)), "status clear-%d", i);
> -		report(check_irq() == (i % 2), "irq-%d", i);
>  	}
>  
>  	report_prefix_pop();
> @@ -440,8 +456,10 @@ static void check_running_counter_wrmsr(void)
>  	report(evt.count < gp_events[1].min, "cntr");
>  
>  	/* clear status before overflow test */
> -	wrmsr(MSR_CORE_PERF_GLOBAL_OVF_CTRL,
> -	      rdmsr(MSR_CORE_PERF_GLOBAL_STATUS));
> +	if (pmu_version() > 1) {

Helper.  Curly braces aren't necessary.

> +		wrmsr(MSR_CORE_PERF_GLOBAL_OVF_CTRL,
> +			rdmsr(MSR_CORE_PERF_GLOBAL_STATUS));
> +	}
>  
>  	start_event(&evt);
>  
> @@ -453,8 +471,11 @@ static void check_running_counter_wrmsr(void)
>  
>  	loop();
>  	stop_event(&evt);
> -	status = rdmsr(MSR_CORE_PERF_GLOBAL_STATUS);
> -	report(status & 1, "status");
> +
> +	if (pmu_version() > 1) {

Helper.

> +		status = rdmsr(MSR_CORE_PERF_GLOBAL_STATUS);
> +		report(status & 1, "status");

Can you opportunistically provide a better message than "status"?

> +	}
>  
>  	report_prefix_pop();
>  }
> @@ -474,8 +495,10 @@ static void check_emulated_instr(void)
>  	};
>  	report_prefix_push("emulated instruction");
>  
> -	wrmsr(MSR_CORE_PERF_GLOBAL_OVF_CTRL,
> -	      rdmsr(MSR_CORE_PERF_GLOBAL_STATUS));
> +	if (pmu_version() > 1) {

Helper, no curly braces.  Ah, IIRC, kernel perf prefers curly braces if the code
spans multiple lines.  KVM and KUT does not.

> +		wrmsr(MSR_CORE_PERF_GLOBAL_OVF_CTRL,
> +			rdmsr(MSR_CORE_PERF_GLOBAL_STATUS));
> +	}
>  
>  	start_event(&brnch_cnt);
>  	start_event(&instr_cnt);
> @@ -509,7 +532,8 @@ static void check_emulated_instr(void)
>  		:
>  		: "eax", "ebx", "ecx", "edx");
>  
> -	wrmsr(MSR_CORE_PERF_GLOBAL_CTRL, 0);
> +	if (pmu_version() > 1)

Helper.

> +		wrmsr(MSR_CORE_PERF_GLOBAL_CTRL, 0);
>  
>  	stop_event(&brnch_cnt);
>  	stop_event(&instr_cnt);
> @@ -520,10 +544,13 @@ static void check_emulated_instr(void)
>  	       "instruction count");
>  	report(brnch_cnt.count - brnch_start >= EXPECTED_BRNCH,
>  	       "branch count");
> -	// Additionally check that those counters overflowed properly.
> -	status = rdmsr(MSR_CORE_PERF_GLOBAL_STATUS);
> -	report(status & 1, "instruction counter overflow");
> -	report(status & 2, "branch counter overflow");
> +
> +	if (pmu_version() > 1) {

Helper?  E.g. if this is a "has architectural PMU".

> +		// Additionally check that those counters overflowed properly.
> +		status = rdmsr(MSR_CORE_PERF_GLOBAL_STATUS);
> +		report(status & 1, "instruction counter overflow");
> +		report(status & 2, "branch counter overflow");
> +	}
>  
>  	report_prefix_pop();
>  }
> @@ -647,12 +674,7 @@ int main(int ac, char **av)
>  	buf = malloc(N*64);
>  
>  	if (!pmu_version()) {
> -		report_skip("No pmu is detected!");
> -		return report_summary();
> -	}
> -
> -	if (pmu_version() == 1) {
> -		report_skip("PMU version 1 is not supported.");
> +		report_skip("No Intel Arch PMU is detected!");
>  		return report_summary();
>  	}
>  
> -- 
> 2.37.2
> 

  parent reply	other threads:[~2022-10-05 22:35 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-08-19 11:09 [kvm-unit-tests PATCH v3 00/13] x86/pmu: Test case optimization, fixes and additions Like Xu
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 01/13] x86/pmu: Introduce __start_event() to drop all of the manual zeroing Like Xu
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 02/13] x86/pmu: Introduce multiple_{one, many}() to improve readability Like Xu
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 03/13] x86/pmu: Reset the expected count of the fixed counter 0 when i386 Like Xu
2022-10-05 22:18   ` Sean Christopherson
2022-10-17  7:30     ` Like Xu
2022-10-21 19:16       ` Sean Christopherson
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 04/13] x86/pmu: Add tests for Intel Processor Event Based Sampling (PEBS) Like Xu
2022-10-05 22:21   ` Sean Christopherson
2022-10-05 22:22     ` Sean Christopherson
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 05/13] x86: create pmu group for quick pmu-scope testing Like Xu
2022-10-05 22:23   ` Sean Christopherson
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 06/13] x86/pmu: Test emulation instructions on full-width counters Like Xu
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 07/13] x86/pmu: Pop up FW prefix to avoid out-of-context propagation Like Xu
2022-10-05 22:25   ` Sean Christopherson
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 08/13] x86/pmu: Add PDCM check before accessing PERF_CAP register Like Xu
2022-10-05 22:28   ` Sean Christopherson
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 09/13] x86/pmu: Report SKIP when testing Intel LBR on AMD platforms Like Xu
2022-10-05 22:29   ` Sean Christopherson
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 10/13] x86/pmu: Update testcases to cover Intel Arch PMU Version 1 Like Xu
2022-09-06  7:15   ` Sandipan Das
2022-09-06 13:28     ` Like Xu
2022-09-06  8:16   ` Sandipan Das
2022-09-06 13:35     ` Like Xu
2022-09-08  8:23       ` Sandipan Das
2022-09-19  7:09         ` Like Xu
2022-10-21  7:32           ` Like Xu
2022-10-25  8:34             ` Sandipan Das
2022-10-05 22:35   ` Sean Christopherson [this message]
2022-10-18  9:32     ` Like Xu
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 11/13] x86/pmu: Refine message when testing PMU on AMD platforms Like Xu
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 12/13] x86/pmu: Add assignment framework for Intel-specific HW resources Like Xu
2022-09-06  7:19   ` Sandipan Das
2022-10-05 22:44   ` Sean Christopherson
2022-10-21  7:21     ` Like Xu
2022-10-21 18:22       ` Sean Christopherson
2022-08-19 11:09 ` [kvm-unit-tests PATCH v3 13/13] x86/pmu: Update testcases to cover AMD PMU Like Xu
2022-09-06  7:32   ` Sandipan Das
2022-10-05 22:48   ` Sean Christopherson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=Yz4GtqyPIMCMsUEl@google.com \
    --to=seanjc@google.com \
    --cc=kvm@vger.kernel.org \
    --cc=like.xu.linux@gmail.com \
    --cc=pbonzini@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).