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

[Xiaoyao Li]

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.

> +        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;