Re: [RFC KERNEL PATCH v2 1/3] xen/pci: Add xen_reset_device_state function

[Stewart Hildebrand <stewart.hildebrand@amd.com>]

On 12/3/23 22:25, Chen, Jiqian wrote:
> Hi Stewart,
> 
> On 2023/11/30 23:03, Stewart Hildebrand wrote:
>> On 11/30/23 02:03, Chen, Jiqian wrote:
>>>
>>> On 2023/11/30 11:46, Stefano Stabellini wrote:
>>>> On Fri, 24 Nov 2023, Jiqian Chen wrote:
>>>>> When device on dom0 side has been reset, the vpci on Xen side
>>>>> won't get notification, so that the cached state in vpci is
>>>>> all out of date with the real device state.
>>>>> To solve that problem, this patch add a function to clear all
>>>>> vpci device state when device is reset on dom0 side.
>>>>>
>>>>> And call that function in pcistub_init_device. Because when
>>>>> we use "pci-assignable-add" to assign a passthrough device in
>>>>> Xen, it will reset passthrough device and the vpci state will
>>>>> out of date, and then device will fail to restore bar state.
>>>>>
>>>>> Signed-off-by: Jiqian Chen <Jiqian.Chen@amd.com>
>>>>> Signed-off-by: Huang Rui <ray.huang@amd.com>
>>>>> ---
>>>>>  drivers/xen/pci.c                  | 12 ++++++++++++
>>>>>  drivers/xen/xen-pciback/pci_stub.c |  3 +++
>>>>>  include/xen/interface/physdev.h    |  2 ++
>>>>>  include/xen/pci.h                  |  6 ++++++
>>>>>  4 files changed, 23 insertions(+)
>>>>>
>>>>> diff --git a/drivers/xen/pci.c b/drivers/xen/pci.c
>>>>> index 72d4e3f193af..e9b30bc09139 100644
>>>>> --- a/drivers/xen/pci.c
>>>>> +++ b/drivers/xen/pci.c
>>>>> @@ -177,6 +177,18 @@ static int xen_remove_device(struct device *dev)
>>>>>  	return r;
>>>>>  }
>>>>>  
>>>>> +int xen_reset_device_state(const struct pci_dev *dev)
>>>>> +{
>>>>> +	struct physdev_pci_device device = {
>>>>> +		.seg = pci_domain_nr(dev->bus),
>>>>> +		.bus = dev->bus->number,
>>>>> +		.devfn = dev->devfn
>>>>> +	};
>>>>> +
>>>>> +	return HYPERVISOR_physdev_op(PHYSDEVOP_pci_device_state_reset, &device);
>>>>> +}
>>>>> +EXPORT_SYMBOL_GPL(xen_reset_device_state);
>>>>> +
>>>>>  static int xen_pci_notifier(struct notifier_block *nb,
>>>>>  			    unsigned long action, void *data)
>>>>>  {
>>>>> diff --git a/drivers/xen/xen-pciback/pci_stub.c b/drivers/xen/xen-pciback/pci_stub.c
>>>>> index e34b623e4b41..5a96b6c66c07 100644
>>>>> --- a/drivers/xen/xen-pciback/pci_stub.c
>>>>> +++ b/drivers/xen/xen-pciback/pci_stub.c
>>>>> @@ -421,6 +421,9 @@ static int pcistub_init_device(struct pci_dev *dev)
>>>>>  	else {
>>>>>  		dev_dbg(&dev->dev, "resetting (FLR, D3, etc) the device\n");
>>>>>  		__pci_reset_function_locked(dev);
>>>>> +		err = xen_reset_device_state(dev);
>>>>> +		if (err)
>>>>> +			goto config_release;
>>>>
>>>> Older versions of Xen won't have the hypercall
>>>> PHYSDEVOP_pci_device_state_reset implemented. I think we should do
>>>> something like:
>>>>
>>>> if (err && xen_pvh_domain())
>>>>     goto config_release;
>>>>
>>>>
>>>> Or even:
>>>>
>>>> if (xen_pvh_domain()) {
>>>>     err = xen_reset_device_state(dev);
>>>>     if (err)
>>>>         goto config_release;
>>>> }
>>>>
>>>> depending on whether we want to call xen_reset_device_state also for PV
>>>> guests or not. I am assuming we don't want to error out on failure such
>>>> as -ENOENT for PV guests.
>>> Yes, only for PVH dom0, I will add the condition in next version. Thank you!
>>
>> We will want to call xen_reset_device_state() for Arm dom0, too, so checking xen_pvh_domain() alone is not sufficient. I suggest instead to check !xen_pv_domain().
> I am not using Arm. But is Arm dom0 not a PVH type dom0?

Apparently not, Arm dom0 appears to be a xen_hvm_domain() from the kernel's perspective.

> 
>>
>>>
>>>>
>>>>
>>>>>  		pci_restore_state(dev);
>>>>>  	}
>>>>>  	/* Now disable the device (this also ensures some private device
>>>>> diff --git a/include/xen/interface/physdev.h b/include/xen/interface/physdev.h
>>>>> index a237af867873..231526f80f6c 100644
>>>>> --- a/include/xen/interface/physdev.h
>>>>> +++ b/include/xen/interface/physdev.h
>>>>> @@ -263,6 +263,8 @@ struct physdev_pci_device {
>>>>>      uint8_t devfn;
>>>>>  };
>>>>>  
>>>>> +#define PHYSDEVOP_pci_device_state_reset     32
>>>>> +
>>>>>  #define PHYSDEVOP_DBGP_RESET_PREPARE    1
>>>>>  #define PHYSDEVOP_DBGP_RESET_DONE       2
>>>>>  
>>>>> diff --git a/include/xen/pci.h b/include/xen/pci.h
>>>>> index b8337cf85fd1..b2e2e856efd6 100644
>>>>> --- a/include/xen/pci.h
>>>>> +++ b/include/xen/pci.h
>>>>> @@ -4,10 +4,16 @@
>>>>>  #define __XEN_PCI_H__
>>>>>  
>>>>>  #if defined(CONFIG_XEN_DOM0)
>>>>> +int xen_reset_device_state(const struct pci_dev *dev);
>>>>>  int xen_find_device_domain_owner(struct pci_dev *dev);
>>>>>  int xen_register_device_domain_owner(struct pci_dev *dev, uint16_t domain);
>>>>>  int xen_unregister_device_domain_owner(struct pci_dev *dev);
>>>>>  #else
>>>>> +static inline int xen_reset_device_state(const struct pci_dev *dev)
>>>>> +{
>>>>> +	return -1;
>>>>> +}
>>>>> +
>>>>>  static inline int xen_find_device_domain_owner(struct pci_dev *dev)
>>>>>  {
>>>>>  	return -1;
>>>>> -- 
>>>>> 2.34.1
>>>>>
>>>
>