Re: [PATCH v5 14/23] iommu: introduce page response function

[Auger Eric <eric.auger@redhat.com>]

Hi Jacob,

On 09/10/2018 07:50 PM, Jacob Pan wrote:
> On Mon, 10 Sep 2018 16:52:24 +0200
> Auger Eric <eric.auger@redhat.com> wrote:
> 
>> Hi Jacob,
>>
> Hi Eric,
> 
> Thanks for the review, please comments inline.
>> On 05/11/2018 10:54 PM, Jacob Pan wrote:
>>> IO page faults can be handled outside IOMMU subsystem. For an
>>> example, when nested translation is turned on and guest owns the
>>> first level page tables, device page request can be forwared  
>> forwarded
>>> to the guest for handling faults. As the page response returns
>>> by the guest, IOMMU driver on the host need to process the  
>> from the guest ...  host needs
>>> response which informs the device and completes the page request
>>> transaction.
>>>
>>> This patch introduces generic API function for page response
>>> passing from the guest or other in-kernel users. The definitions of
>>> the generic data is based on PCI ATS specification not limited to
>>> any vendor.
>>>
>>> Signed-off-by: Jean-Philippe Brucker <jean-philippe.brucker@arm.com>
>>> Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
>>> Link: https://lkml.org/lkml/2017/12/7/1725
>>> ---
>>>  drivers/iommu/iommu.c | 45
>>> +++++++++++++++++++++++++++++++++++++++++++++ include/linux/iommu.h
>>> | 43 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed,
>>> 88 insertions(+)
>>>
>>> diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
>>> index b3f9daf..02fed3e 100644
>>> --- a/drivers/iommu/iommu.c
>>> +++ b/drivers/iommu/iommu.c
>>> @@ -1533,6 +1533,51 @@ int iommu_sva_invalidate(struct iommu_domain
>>> *domain, }
>>>  EXPORT_SYMBOL_GPL(iommu_sva_invalidate);
>>>  
>>> +int iommu_page_response(struct device *dev,
>>> +			struct page_response_msg *msg)
>>> +{
>>> +	struct iommu_param *param = dev->iommu_param;
>>> +	int ret = -EINVAL;
>>> +	struct iommu_fault_event *evt;
>>> +	struct iommu_domain *domain =
>>> iommu_get_domain_for_dev(dev); +
>>> +	if (!domain || !domain->ops->page_response)
>>> +		return -ENODEV;
>>> +
>>> +	/*
>>> +	 * Device iommu_param should have been allocated when
>>> device is
>>> +	 * added to its iommu_group.
>>> +	 */
>>> +	if (!param || !param->fault_param)
>>> +		return -EINVAL;
>>> +
>>> +	/* Only send response if there is a fault report pending */
>>> +	mutex_lock(&param->fault_param->lock);
>>> +	if (list_empty(&param->fault_param->faults)) {
>>> +		pr_warn("no pending PRQ, drop response\n");
>>> +		goto done_unlock;
>>> +	}
>>> +	/*
>>> +	 * Check if we have a matching page request pending to
>>> respond,
>>> +	 * otherwise return -EINVAL
>>> +	 */
>>> +	list_for_each_entry(evt, &param->fault_param->faults,
>>> list) {
>>> +		if (evt->pasid == msg->pasid &&
>>> +		    msg->page_req_group_id ==
>>> evt->page_req_group_id) {
>>> +			msg->private_data = evt->iommu_private;
>>> +			ret = domain->ops->page_response(dev, msg);
>>> +			list_del(&evt->list);  
>> don't you need a list_for_each_entry_safe?
> why? I am here exiting the loop.
>>> +			kfree(evt);
>>> +			break;
Ah OK I missed the break. If you delete a single entry per page response
it is OK then. sorry for the noise.
>>> +		}
>>> +	}
>>> +
>>> +done_unlock:
>>> +	mutex_unlock(&param->fault_param->lock);
>>> +	return ret;
>>> +}
>>> +EXPORT_SYMBOL_GPL(iommu_page_response);
>>> +
>>>  static void __iommu_detach_device(struct iommu_domain *domain,
>>>  				  struct device *dev)
>>>  {
>>> diff --git a/include/linux/iommu.h b/include/linux/iommu.h
>>> index b3312ee..722b90f 100644
>>> --- a/include/linux/iommu.h
>>> +++ b/include/linux/iommu.h
>>> @@ -163,6 +163,41 @@ struct iommu_resv_region {
>>>  #ifdef CONFIG_IOMMU_API
>>>  
>>>  /**
>>> + * enum page_response_code - Return status of fault handlers,
>>> telling the IOMMU
>>> + * driver how to proceed with the fault.
>>> + *
>>> + * @IOMMU_PAGE_RESP_SUCCESS: Fault has been handled and the page
>>> tables
>>> + *	populated, retry the access. This is "Success" in PCI
>>> PRI.
>>> + * @IOMMU_PAGE_RESP_FAILURE: General error. Drop all subsequent
>>> faults from
>>> + *	this device if possible. This is "Response Failure" in
>>> PCI PRI.
>>> + * @IOMMU_PAGE_RESP_INVALID: Could not handle this fault, don't
>>> retry the
>>> + *	access. This is "Invalid Request" in PCI PRI.
>>> + */
>>> +enum page_response_code {
>>> +	IOMMU_PAGE_RESP_SUCCESS = 0,
>>> +	IOMMU_PAGE_RESP_INVALID,
>>> +	IOMMU_PAGE_RESP_FAILURE,
>>> +};
>>> +
>>> +/**
>>> + * Generic page response information based on PCI ATS and PASID
>>> spec.
>>> + * @addr: servicing page address
>>> + * @pasid: contains process address space ID
>>> + * @resp_code: response code  
>> nit: @pasid_present doc missing although quite obvious
>>> + * @page_req_group_id: page request group index
>>> + * @private_data: uniquely identify device-specific private data
>>> for an
>>> + *                individual page response
>>> + */
>>> +struct page_response_msg {
>>> +	u64 addr;
>>> +	u32 pasid;
>>> +	enum page_response_code resp_code;
>>> +	u32 pasid_present:1;
>>> +	u32 page_req_group_id;
>>> +	u64 private_data;
>>> +};  
>> Doesn't it need to be part of iommu uapi header since the virtualizer
>> will pass the response through VFIO?
>>
> Right, that has been the same feedback from others as well. I am moving
> it to uapi in the next rev.
>> As mentioned in previous discussion this is really PRI related and
>> does not really fit unrecoverable fault reporting. To me we should
>> clarify if this API targets both use cases or only the PRI response
>> use case.
> Yes, I should clarify this is for PRI only. It is little bit asymmetric
> in that per IOMMU device fault reporting covers both unrecoverable
> faults and PRI, but only PRI needs page response.
OK. Still unrecoverable errors need a "read" API as the virtualizer may
inject them into a guest. The fault handler may signal an eventfd and
the userspace handler needs to retrieve the pending fault event(s).
> 
>> Also in the implementation we check pasid and PRGindex. As
>> mentionned by Jean-Philippe, unrecoverable "traditional" faults do
>> not require to manage a list in the iommu subsystem.

>>
> I am not sure if that is a question. We support PRI with PASID only.
> We keep the group ID for page responses.
As I was trying to reuse this API for unrecoverable errors for SMMU
stage1, (unrelated to PRI management), the check of pasid and PRGindex
looked very PRI specific.
>> Have you considered using a kfifo instead of a list to manage the
>> pending PRI requests?
>>
> No, I will look into it. But we may need too traverse the list in case
> of exceptions. e.g. dropping some pending requests if device faults or
> process/vm terminates.
Yes thinking more about it the kfifo does not seem to be adapted to your
needs. Also I think the PRI requests may be sent out of order (?). Kfifo
looks more adapted to unrecoverable errors.

Thanks

Eric
> 
>> Thanks
>>
>> Eric
>>> +
>>> +/**
>>>   * struct iommu_ops - iommu ops and capabilities
>>>   * @capable: check capability
>>>   * @domain_alloc: allocate iommu domain
>>> @@ -195,6 +230,7 @@ struct iommu_resv_region {
>>>   * @bind_pasid_table: bind pasid table pointer for guest SVM
>>>   * @unbind_pasid_table: unbind pasid table pointer and restore
>>> defaults
>>>   * @sva_invalidate: invalidate translation caches of shared
>>> virtual address
>>> + * @page_response: handle page request response
>>>   */
>>>  struct iommu_ops {
>>>  	bool (*capable)(enum iommu_cap);
>>> @@ -250,6 +286,7 @@ struct iommu_ops {
>>>  				struct device *dev);
>>>  	int (*sva_invalidate)(struct iommu_domain *domain,
>>>  		struct device *dev, struct tlb_invalidate_info
>>> *inv_info);
>>> +	int (*page_response)(struct device *dev, struct
>>> page_response_msg *msg); 
>>>  	unsigned long pgsize_bitmap;
>>>  };
>>> @@ -470,6 +507,7 @@ extern int
>>> iommu_unregister_device_fault_handler(struct device *dev); 
>>>  extern int iommu_report_device_fault(struct device *dev, struct
>>> iommu_fault_event *evt); 
>>> +extern int iommu_page_response(struct device *dev, struct
>>> page_response_msg *msg); extern int iommu_group_id(struct
>>> iommu_group *group); extern struct iommu_group
>>> *iommu_group_get_for_dev(struct device *dev); extern struct
>>> iommu_domain *iommu_group_default_domain(struct iommu_group *); @@
>>> -758,6 +796,11 @@ static inline int
>>> iommu_report_device_fault(struct device *dev, struct iommu_fau
>>> return -ENODEV; } 
>>> +static inline int iommu_page_response(struct device *dev, struct
>>> page_response_msg *msg) +{
>>> +	return -ENODEV;
>>> +}
>>> +
>>>  static inline int iommu_group_id(struct iommu_group *group)
>>>  {
>>>  	return -ENODEV;
>>>   
> 
> [Jacob Pan]
>