Re: [RFC PATCH] i386: Add ratelimit for bus locks acquired in guest

[Chenyi Qiang <chenyi.qiang@intel.com>]

On 3/19/2021 9:23 AM, Xiaoyao Li wrote:
> On 3/17/2021 4:47 PM, Chenyi Qiang wrote:
> [...]
>>   MemTxAttrs kvm_arch_post_run(CPUState *cpu, struct kvm_run *run)
>>   {
>>       X86CPU *x86_cpu = X86_CPU(cpu);
>> @@ -4236,6 +4271,11 @@ MemTxAttrs kvm_arch_post_run(CPUState *cpu, 
>> struct kvm_run *run)
>>       } else {
>>           env->eflags &= ~IF_MASK;
>>       }
>> +    if (run->flags & KVM_RUN_X86_BUS_LOCK) {
>> +        kvm_cpu_synchronize_state(cpu);
>> +        warn_report("bus lock detected at rip: 0x%lx", env->eip);
> 
> Chenyi,
> 
> Let's drop the eip here since QEMU has no idea whether it points to the 
> next instruction or the exact instruction acquires bus lock.
> 

Fair enough.

>> +        kvm_rate_limit_on_bus_lock();
>> +    }
>>       /* We need to protect the apic state against concurrent accesses 
>> from
>>        * different threads in case the userspace irqchip is used. */
>> @@ -4594,6 +4634,10 @@ int kvm_arch_handle_exit(CPUState *cs, struct 
>> kvm_run *run)
>>           ioapic_eoi_broadcast(run->eoi.vector);
>>           ret = 0;
>>           break;
>> +    case KVM_EXIT_X86_BUS_LOCK:
>> +        /* already handled in kvm_arch_post_run */
>> +        ret = 0;
>> +        break;
>>       default:
>>           fprintf(stderr, "KVM: unknown exit reason %d\n", 
>> run->exit_reason);
>>           ret = -1;
>>
>