Re: [PATCH 1/1] virtio: add driver_override support

[Max Gurtovoy <mgurtovoy@nvidia.com>]

On 09/12/2025 15:59, Stefan Hajnoczi wrote:
> On Mon, Dec 08, 2025 at 06:18:14PM +0200, Max Gurtovoy wrote:
>> On 08/12/2025 17:54, Stefan Hajnoczi wrote:
>>> On Mon, Dec 08, 2025 at 04:39:25PM +0200, Max Gurtovoy wrote:
>>>> Add support for the 'driver_override' attribute to Virtio devices. This
>>>> allows users to control which Virtio bus driver binds to a given Virtio
>>>> device.
>>>>
>>>> If 'driver_override' is not set, the existing behavior is preserved and
>>>> devices will continue to auto-bind to the first matching Virtio bus
>>>> driver.
>>>>
>>>> Signed-off-by: Avraham Evdaev <aevdaev@nvidia.com>
>>>> Signed-off-by: Max Gurtovoy <mgurtovoy@nvidia.com>
>>>> ---
>>>>    drivers/virtio/virtio.c | 34 ++++++++++++++++++++++++++++++++++
>>>>    include/linux/virtio.h  |  4 ++++
>>>>    2 files changed, 38 insertions(+)
>>> What is the use case? Is there something missing in existing drivers
>>> that cannot be added to them by extending the code?
>>>
>>> Stefan
>> The main goal is to align the virtio bus with the flexibility that already exists for other buses in the Linux device model.
>> On buses such as PCI and vDPA, it is possible to override or replace the default driver for a given device and bind it to an alternative driver.
>> Allowing the selection of which driver is bound to a device can be useful in both development and production environments.
> There are use cases for PCI and vDPA, like binding vfio_pci to a device
> that would normally use its device-specific driver.
>
> For VIRTIO, I'm not sure what the use case would be. Is there another
> driver that could bind to virtio-net devices, for example?
Yes.
>
> Giving a concrete example would be useful to help people understand why
> this is necessary.

See example in my v2.

-Max

>
> Stefan