qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Xiaoyao Li <xiaoyao.li@intel.com>
To: dongli.zhang@oracle.com, qemu-devel@nongnu.org, kvm@vger.kernel.org
Cc: pbonzini@redhat.com, zhao1.liu@intel.com, mtosatti@redhat.com,
	sandipan.das@amd.com, babu.moger@amd.com, likexu@tencent.com,
	like.xu.linux@gmail.com, zhenyuw@linux.intel.com, groug@kaod.org,
	khorenko@virtuozzo.com, alexander.ivanov@virtuozzo.com,
	den@virtuozzo.com, davydov-max@yandex-team.ru,
	dapeng1.mi@linux.intel.com, joe.jin@oracle.com
Subject: Re: [PATCH v2 02/10] target/i386: disable PERFCORE when "-pmu" is configured
Date: Tue, 4 Mar 2025 14:11:27 +0800	[thread overview]
Message-ID: <addd4bf3-af4c-446d-b586-62c89ab37d26@intel.com> (raw)
In-Reply-To: <0dcdd9c1-35e5-4cee-be0b-59113e01e73c@oracle.com>

On 3/4/2025 2:45 AM, dongli.zhang@oracle.com wrote:
> Hi Xiaoyao,
> 
> On 3/2/25 5:59 PM, Xiaoyao Li wrote:
>> On 3/3/2025 6:00 AM, Dongli Zhang wrote:
>>> Currently, AMD PMU support isn't determined based on CPUID, that is, the
>>> "-pmu" option does not fully disable KVM AMD PMU virtualization.
>>>
>>> To minimize AMD PMU features, remove PERFCORE when "-pmu" is configured.
>>>
>>> To completely disable AMD PMU virtualization will be implemented via
>>> KVM_CAP_PMU_CAPABILITY in upcoming patches.
>>>
>>> As a reminder, neither CPUID_EXT3_PERFCORE nor
>>> CPUID_8000_0022_EAX_PERFMON_V2 is removed from env->features[] when "-pmu"
>>> is configured. Developers should query whether they are supported via
>>> cpu_x86_cpuid() rather than relying on env->features[] in future patches.
>>
>> I don't think it is the correct direction to go.
>>
>> env->features[] should be finalized before cpu_x86_cpuid() and env-
>>> features[] needs to be able to be exposed to guest directly. This ensures
>> guest and QEMU have the same view of CPUIDs and it simplifies things.
>>
>> We can adjust env->features[] by filtering all PMU related CPUIDs based on
>> cpu->enable_pmu in x86_cpu_realizefn().
> 
> Thank you very much for suggestion.
> 
> I see  code like below in x86_cpu_realizefn() to edit env->features[].
> 
> 7982     /* On AMD CPUs, some CPUID[8000_0001].EDX bits must match the bits on
> 7983      * CPUID[1].EDX.
> 7984      */
> 7985     if (IS_AMD_CPU(env)) {
> 7986         env->features[FEAT_8000_0001_EDX] &= ~CPUID_EXT2_AMD_ALIASES;
> 7987         env->features[FEAT_8000_0001_EDX] |= (env->features[FEAT_1_EDX]
> 7988            & CPUID_EXT2_AMD_ALIASES);
> 7989     }
> 
> I may do something similar to them for CPUID_EXT3_PERFCORE and
> CPUID_8000_0022_EAX_PERFMON_V2.

I just sent a series for CPUID_EXT_PDCM[1]. I think you can put 
CPUID_EXT3_PERFCORE and CPUID_8000_0022_EAX_PERFMON_V2 at the same place.

[1] 
https://lore.kernel.org/qemu-devel/20250304052450.465445-1-xiaoyao.li@intel.com/T/#m31c6777131b6361d7c3af22b09532bdc785dbc06

> Dongli Zhang
> 
> 
> 
>>
>>> Suggested-by: Zhao Liu <zhao1.liu@intel.com>
>>> Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
>>> ---
>>>    target/i386/cpu.c | 4 ++++
>>>    1 file changed, 4 insertions(+)
>>>
>>> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
>>> index b6d6167910..61a671028a 100644
>>> --- a/target/i386/cpu.c
>>> +++ b/target/i386/cpu.c
>>> @@ -7115,6 +7115,10 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t
>>> index, uint32_t count,
>>>                !(env->hflags & HF_LMA_MASK)) {
>>>                *edx &= ~CPUID_EXT2_SYSCALL;
>>>            }
>>> +
>>> +        if (kvm_enabled() && IS_AMD_CPU(env) && !cpu->enable_pmu) {
>>> +            *ecx &= ~CPUID_EXT3_PERFCORE;
>>> +        }
>>>            break;
>>>        case 0x80000002:
>>>        case 0x80000003:
>>
> 



  reply	other threads:[~2025-03-04  6:12 UTC|newest]

Thread overview: 63+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-02 22:00 [PATCH v2 00/10] target/i386/kvm/pmu: PMU Enhancement, Bugfix and Cleanup Dongli Zhang
2025-03-02 22:00 ` [PATCH v2 01/10] target/i386: disable PerfMonV2 when PERFCORE unavailable Dongli Zhang
2025-03-04 14:40   ` Xiaoyao Li
2025-03-04 22:53     ` dongli.zhang
2025-03-05  1:38       ` Xiaoyao Li
2025-03-05 14:20   ` Zhao Liu
2025-03-07  7:24   ` Sandipan Das
2025-03-02 22:00 ` [PATCH v2 02/10] target/i386: disable PERFCORE when "-pmu" is configured Dongli Zhang
2025-03-03  1:59   ` Xiaoyao Li
2025-03-03 18:45     ` dongli.zhang
2025-03-04  6:11       ` Xiaoyao Li [this message]
2025-03-06 16:50   ` Zhao Liu
2025-03-06 17:47     ` dongli.zhang
2025-03-07  7:41       ` Zhao Liu
2025-03-02 22:00 ` [PATCH v2 03/10] [DO NOT MERGE] kvm: Introduce kvm_arch_pre_create_vcpu() Dongli Zhang
2025-03-05 14:46   ` Zhao Liu
2025-03-05 21:53     ` dongli.zhang
2025-03-07  7:52       ` Zhao Liu
2025-03-07  8:40         ` Xiaoyao Li
2025-03-02 22:00 ` [PATCH v2 04/10] target/i386/kvm: set KVM_PMU_CAP_DISABLE if "-pmu" is configured Dongli Zhang
2025-03-04  7:59   ` Xiaoyao Li
2025-03-05  1:22     ` Sean Christopherson
2025-03-05  1:35       ` Xiaoyao Li
2025-03-05 14:41     ` Zhao Liu
2025-03-05 20:13       ` dongli.zhang
2025-03-05 14:44   ` Zhao Liu
2025-03-02 22:00 ` [PATCH v2 05/10] target/i386/kvm: extract unrelated code out of kvm_x86_build_cpuid() Dongli Zhang
2025-03-05  7:03   ` Mi, Dapeng
2025-03-07  9:15   ` Zhao Liu
2025-03-07 22:47     ` Dongli Zhang
2025-03-10  3:55       ` Zhao Liu
2025-03-02 22:00 ` [PATCH v2 06/10] target/i386/kvm: rename architectural PMU variables Dongli Zhang
2025-03-05  7:07   ` Mi, Dapeng
2025-03-07  9:19   ` Zhao Liu
2025-03-07 22:49     ` Dongli Zhang
2025-03-02 22:00 ` [PATCH v2 07/10] target/i386/kvm: query kvm.enable_pmu parameter Dongli Zhang
2025-03-10  6:14   ` Zhao Liu
2025-03-10 15:41     ` Dongli Zhang
2025-03-10 16:49     ` Dongli Zhang
2025-03-02 22:00 ` [PATCH v2 08/10] target/i386/kvm: reset AMD PMU registers during VM reset Dongli Zhang
2025-03-05  7:33   ` Mi, Dapeng
2025-03-05 11:41   ` Francesco Lavra
2025-03-05 19:05     ` dongli.zhang
2025-03-07  7:38   ` Sandipan Das
2025-03-10  7:47   ` Zhao Liu
2025-03-10 16:39     ` Dongli Zhang
2025-03-11 13:51       ` Zhao Liu
2025-03-11 19:52         ` Dongli Zhang
2025-03-12  8:30           ` Zhao Liu
2025-03-12 22:17             ` Dongli Zhang
2025-03-28  6:29   ` ewanhai
2025-03-28 16:42     ` Dongli Zhang
2025-03-31  3:55       ` ewanhai
2025-03-31 19:16         ` Dongli Zhang
2025-04-01  3:35           ` Ewan Hai
2025-04-07  8:51             ` Zhao Liu
2025-04-07  9:33               ` Ewan Hai
2025-04-16  8:17                 ` Mi, Dapeng
2025-03-02 22:00 ` [PATCH v2 09/10] target/i386/kvm: support perfmon-v2 for reset Dongli Zhang
2025-03-02 22:00 ` [PATCH v2 10/10] target/i386/kvm: don't stop Intel PMU counters Dongli Zhang
2025-03-05  7:35   ` Mi, Dapeng
2025-03-05 19:00     ` dongli.zhang
2025-03-06  1:38       ` Mi, Dapeng

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=addd4bf3-af4c-446d-b586-62c89ab37d26@intel.com \
    --to=xiaoyao.li@intel.com \
    --cc=alexander.ivanov@virtuozzo.com \
    --cc=babu.moger@amd.com \
    --cc=dapeng1.mi@linux.intel.com \
    --cc=davydov-max@yandex-team.ru \
    --cc=den@virtuozzo.com \
    --cc=dongli.zhang@oracle.com \
    --cc=groug@kaod.org \
    --cc=joe.jin@oracle.com \
    --cc=khorenko@virtuozzo.com \
    --cc=kvm@vger.kernel.org \
    --cc=like.xu.linux@gmail.com \
    --cc=likexu@tencent.com \
    --cc=mtosatti@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=sandipan.das@amd.com \
    --cc=zhao1.liu@intel.com \
    --cc=zhenyuw@linux.intel.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).