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().
Suggested-by: Zhao Liu <zhao1.liu@xxxxxxxxx>
Signed-off-by: Dongli Zhang <dongli.zhang@xxxxxxxxxx>
---
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:
