diff mbox series

[1/7] target/i386: disable PerfMonV2 when PERFCORE unavailable

Message ID 20241104094119.4131-2-dongli.zhang@oracle.com (mailing list archive)
State New
Headers show
Series target/i386/kvm/pmu: Enhancement, Bugfix and Cleanup | expand

Commit Message

Dongli Zhang Nov. 4, 2024, 9:40 a.m. UTC
When the PERFCORE is disabled with "-cpu host,-perfctr-core", it is
reflected in in guest dmesg.

[    0.285136] Performance Events: AMD PMU driver.

However, the guest cpuid indicates the PerfMonV2 is still available.

CPU:
   Extended Performance Monitoring and Debugging (0x80000022):
      AMD performance monitoring V2         = true
      AMD LBR V2                            = false
      AMD LBR stack & PMC freezing          = false
      number of core perf ctrs              = 0x6 (6)
      number of LBR stack entries           = 0x0 (0)
      number of avail Northbridge perf ctrs = 0x0 (0)
      number of available UMC PMCs          = 0x0 (0)
      active UMCs bitmask                   = 0x0

Disable PerfMonV2 in cpuid when PERFCORE is disabled.

Fixes: 209b0ac12074 ("target/i386: Add PerfMonV2 feature bit")
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
---
 target/i386/cpu.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Zhao Liu Nov. 6, 2024, 3:54 a.m. UTC | #1
Hi Dongli,

On Mon, Nov 04, 2024 at 01:40:16AM -0800, Dongli Zhang wrote:
> Date: Mon,  4 Nov 2024 01:40:16 -0800
> From: Dongli Zhang <dongli.zhang@oracle.com>
> Subject: [PATCH 1/7] target/i386: disable PerfMonV2 when PERFCORE
>  unavailable
> X-Mailer: git-send-email 2.43.5
> 
> When the PERFCORE is disabled with "-cpu host,-perfctr-core", it is
> reflected in in guest dmesg.
> 
> [    0.285136] Performance Events: AMD PMU driver.
> 
> However, the guest cpuid indicates the PerfMonV2 is still available.
> 
> CPU:
>    Extended Performance Monitoring and Debugging (0x80000022):
>       AMD performance monitoring V2         = true
>       AMD LBR V2                            = false
>       AMD LBR stack & PMC freezing          = false
>       number of core perf ctrs              = 0x6 (6)
>       number of LBR stack entries           = 0x0 (0)
>       number of avail Northbridge perf ctrs = 0x0 (0)
>       number of available UMC PMCs          = 0x0 (0)
>       active UMCs bitmask                   = 0x0
> 
> Disable PerfMonV2 in cpuid when PERFCORE is disabled.
> 
> Fixes: 209b0ac12074 ("target/i386: Add PerfMonV2 feature bit")
> Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
> ---
>  target/i386/cpu.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> index 3baa95481f..4490a7a8d6 100644
> --- a/target/i386/cpu.c
> +++ b/target/i386/cpu.c
> @@ -7103,6 +7103,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
>          *eax = *ebx = *ecx = *edx = 0;
>          /* AMD Extended Performance Monitoring and Debug */
>          if (kvm_enabled() && cpu->enable_pmu &&
> +            (env->features[FEAT_8000_0001_ECX] & CPUID_EXT3_PERFCORE) &&
>              (env->features[FEAT_8000_0022_EAX] & CPUID_8000_0022_EAX_PERFMON_V2)) {
>              *eax |= CPUID_8000_0022_EAX_PERFMON_V2;
>              *ebx |= kvm_arch_get_supported_cpuid(cs->kvm_state, index, count,

You can define dependency like this:

diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index 3baa95481fbc..99c69ec9f369 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -1803,6 +1803,10 @@ static FeatureDep feature_dependencies[] = {
         .from = { FEAT_7_1_EDX,             CPUID_7_1_EDX_AVX10 },
         .to = { FEAT_24_0_EBX,              ~0ull },
     },
+    {
+        .from = { FEAT_8000_0001_ECX,       CPUID_EXT3_PERFCORE },
+        .to = { FEAT_8000_0022_EAX,         CPUID_8000_0022_EAX_PERFMON_V2 }
+    }
 };

 typedef struct X86RegisterInfo32 {

---
Does this meet your needs?

Regards,
Zhao
Dongli Zhang Nov. 7, 2024, 12:29 a.m. UTC | #2
Hi Zhao,


> 
> You can define dependency like this:
> 
> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> index 3baa95481fbc..99c69ec9f369 100644
> --- a/target/i386/cpu.c
> +++ b/target/i386/cpu.c
> @@ -1803,6 +1803,10 @@ static FeatureDep feature_dependencies[] = {
>          .from = { FEAT_7_1_EDX,             CPUID_7_1_EDX_AVX10 },
>          .to = { FEAT_24_0_EBX,              ~0ull },
>      },
> +    {
> +        .from = { FEAT_8000_0001_ECX,       CPUID_EXT3_PERFCORE },
> +        .to = { FEAT_8000_0022_EAX,         CPUID_8000_0022_EAX_PERFMON_V2 }
> +    }
>  };
> 
>  typedef struct X86RegisterInfo32 {
> 
> ---
> Does this meet your needs?
> 

Thank you very much for the suggestion.

Yes, this works. The PERFCORE is a prerequisite of PERFMON_V2 (according to
Linux kernel source code).

1403 static int __init amd_core_pmu_init(void)
1404 {
1405         union cpuid_0x80000022_ebx ebx;
1406         u64 even_ctr_mask = 0ULL;
1407         int i;
1408
1409         if (!boot_cpu_has(X86_FEATURE_PERFCTR_CORE))
1410                 return 0;


If you don't mind, I will send the v2 with your Suggested-by.

Thank you very much!

Dongli Zhang
Zhao Liu Nov. 7, 2024, 7:57 a.m. UTC | #3
> Thank you very much for the suggestion.
> 
> Yes, this works. The PERFCORE is a prerequisite of PERFMON_V2 (according to
> Linux kernel source code).
> 
> 1403 static int __init amd_core_pmu_init(void)
> 1404 {
> 1405         union cpuid_0x80000022_ebx ebx;
> 1406         u64 even_ctr_mask = 0ULL;
> 1407         int i;
> 1408
> 1409         if (!boot_cpu_has(X86_FEATURE_PERFCTR_CORE))
> 1410                 return 0;
> 
> 
> If you don't mind, I will send the v2 with your Suggested-by.
> 
> Thank you very much!

Sure, welcome! :)

-Zhao

> Dongli Zhang
diff mbox series

Patch

diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index 3baa95481f..4490a7a8d6 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -7103,6 +7103,7 @@  void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
         *eax = *ebx = *ecx = *edx = 0;
         /* AMD Extended Performance Monitoring and Debug */
         if (kvm_enabled() && cpu->enable_pmu &&
+            (env->features[FEAT_8000_0001_ECX] & CPUID_EXT3_PERFCORE) &&
             (env->features[FEAT_8000_0022_EAX] & CPUID_8000_0022_EAX_PERFMON_V2)) {
             *eax |= CPUID_8000_0022_EAX_PERFMON_V2;
             *ebx |= kvm_arch_get_supported_cpuid(cs->kvm_state, index, count,