Re: [PATCH] KVM/Eventfd: Avoid crash when assign and deassign same eventfd in parallel.

[Paolo Bonzini]

On 19/12/2017 07:35, Lan Tianyu wrote:
> On 2017å12æ18æ 16:50, Paolo Bonzini wrote:
>> On 18/12/2017 09:30, David Hildenbrand wrote:
>>> The ugly thing in kvm_irqfd_assign() is that we access irqfd without
>>> holding a lock. I think that should rather be fixed than working around
>>> that issue. (e.g. lock() -> lookup again -> verify still in list ->
>>> unlock())
>>
>> I wonder if it's even simpler:
>>
>> diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
>> index f2ac53ab8243..17ed298bd66f 100644
>> --- a/virt/kvm/eventfd.c
>> +++ b/virt/kvm/eventfd.c
>> @@ -387,7 +387,6 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
>>  
>>  	idx = srcu_read_lock(&kvm->irq_srcu);
>>  	irqfd_update(kvm, irqfd);
>> -	srcu_read_unlock(&kvm->irq_srcu, idx);
>>  
>>  	list_add_tail(&irqfd->list, &kvm->irqfds.items);
>>  
>> @@ -420,10 +419,12 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
>>  				irqfd->consumer.token, ret);
>>  	}
>>  #endif
>> +	srcu_read_unlock(&kvm->irq_srcu, idx);
>>  
>>  	return 0;
>>  
>>  fail:
>> +	/* irq_srcu is *not* held here.  */
>>  	if (irqfd->resampler)
>>  		irqfd_resampler_shutdown(irqfd);
>>  
>>
> 
> Hi Paolo:
> 	This patch still can't prevent from freeing struct irqfd in
> irq_shutdown() during assign irqfd. Once irqfd is added to
> kvm->irqfds.items list, deassign path can get irqfd and free it.

You're right, that also has to be protected by SRCU.  So a new bool is
needed (probably "active" more than "initialized") in order to replace
list_del_init with list_del_rcu.

Paolo