Re: [RFC KERNEL PATCH v2 2/3] xen/pvh: Unmask irq for passthrough device in PVH dom0

["Chen, Jiqian" <Jiqian.Chen@xxxxxxx>]

Hi Thomas Gleixner,

Thank you for review, and you are right, it seems more like a XEN internal 
issue. We are discussing it and maybe will fix it in Xen code next version.

On 2023/12/4 16:13, Thomas Gleixner wrote:
> On Fri, Nov 24 2023 at 18:31, Jiqian Chen wrote:
>> diff --git a/drivers/xen/xen-pciback/pci_stub.c 
>> b/drivers/xen/xen-pciback/pci_stub.c
>> index 5a96b6c66c07..b83d02bcc76c 100644
>> --- a/drivers/xen/xen-pciback/pci_stub.c
>> +++ b/drivers/xen/xen-pciback/pci_stub.c
>> @@ -357,6 +357,7 @@ static int pcistub_match(struct pci_dev *dev)
>>  static int pcistub_init_device(struct pci_dev *dev)
>>  {
>>      struct xen_pcibk_dev_data *dev_data;
>> +    struct irq_desc *desc = NULL;
>>      int err = 0;
>>  
>>      dev_dbg(&dev->dev, "initializing...\n");
>> @@ -399,6 +400,12 @@ static int pcistub_init_device(struct pci_dev *dev)
>>      if (err)
>>              goto config_release;
>>  
>> +    if (xen_initial_domain() && xen_pvh_domain()) {
>> +            if (dev->irq <= 0 || !(desc = irq_to_desc(dev->irq)))
> 
> Driver code has absolutely no business to access irq_desc.
> 
>> +                    goto config_release;
>> +            unmask_irq(desc);
> 
> Or to invoke any internal function.
> 
>> --- a/kernel/irq/chip.c
>> +++ b/kernel/irq/chip.c
>> @@ -439,6 +439,7 @@ void unmask_irq(struct irq_desc *desc)
>>              irq_state_clr_masked(desc);
>>      }
>>  }
>> +EXPORT_SYMBOL_GPL(unmask_irq);
> 
> Not going to happen.
> 
>> --- a/kernel/irq/irqdesc.c
>> +++ b/kernel/irq/irqdesc.c
>> @@ -380,7 +380,7 @@ struct irq_desc *irq_to_desc(unsigned int irq)
>>  {
>>      return mtree_load(&sparse_irqs, irq);
>>  }
>> -#ifdef CONFIG_KVM_BOOK3S_64_HV_MODULE
>> +#if defined CONFIG_KVM_BOOK3S_64_HV_MODULE || defined CONFIG_XEN_PVH
> 
> Neither that.
> 
> This all smells badly like a XEN internal issue and we are not going to
> hack around it by exposing interrupt internals.
> 
> Thanks,
> 
>         tglx

-- 
Best regards,
Jiqian Chen.