Re: [PATCH] KVM: x86: fix missed hardware breakpoints

[Paolo Bonzini]

On 26/02/2016 11:42, Xiao Guangrong wrote:
>>
>> +    vcpu->arch.switch_db_regs |= KVM_DEBUGREG_RELOAD;
> 
> Er, i do not understand how it works. The BP is enabled in this test case so
> the debug registers are always reloaded before entering guest as
> KVM_DEBUGREG_BP_ENABLED bit is always set on switch_db_regs. What did i
> miss?
> 
> Another impact of this fix is when vcpu is rescheduled we need to always
> reload debug registers even if guest does not enable it, it is really needed?

Hi,

I have looked further at the bug and the issue is that the lazy debug
register optimization doesn't call kvm_update_dr7 and thus does not set
KVM_DEBUGREG_BP_ENABLED.  I will post a better patch shortly.  However,
I still think this one is simpler to have in stable kernel releases,
because it doesn't have any dependencies.

Paolo