Re: [PATCH v3 2/5] vfio/pci: Add an error handler callback

[Markus Armbruster <armbru@redhat.com>]

Farhan Ali <alifm@linux.ibm.com> writes:

> On 9/30/2025 9:52 PM, Markus Armbruster wrote:
>> Farhan Ali <alifm@linux.ibm.com> writes:
>>
>>> On 9/30/2025 2:20 AM, Markus Armbruster wrote:
>>>> Farhan Ali <alifm@linux.ibm.com> writes:
>>>>
>>>>> On 9/27/2025 12:05 AM, Cédric Le Goater wrote:
>>>>>> On 9/27/25 07:59, Markus Armbruster wrote:
>>
>> [...]
>>
>>>>>>> * Is the comment still accurate?
>>>>>
>>>>> This comment would still apply for vfio-pci devices on other architectures except for s390x. We are trying to change this behavior for s390x.
>>>
>>>> The comment is about things that should be done to handle the error.
>>>> Would these things be done here, or in a suitable ->err_handler()?
>>>
>>> Ideally in the err_handler(). And for s390x we try do what the comment mentions, which is inject the error into the guest through s390x architecture specific mechanism. I can remove the comment block.
>>
>> Well, if there's stuff left to do, a comment outlining it is desirable.
>> If I understand you correctly, then the one we have is no longer
>> accurate.  Could you update it, so it is?
>>
>> [...]
>>
> How about something like this?
>
> We can retrieve the error details and decide what action needs to be taken in err_handler(). One of the actions could be to pass the error to the guest and have the guest driver recover from the error. This requires that PCIe capabilities be exposed to the guest. If err_handler() is not implemented/fails, we just terminate the guest to contain the error.

Looks good to me.  I lack the expertise to really judge.