On Sun, Oct 24, 2021 at 4:40 PM Javier Martinez Canillas javierm@redhat.com wrote:
Hello Thomas,
Thanks a lot for your feedback.
On 10/22/21 21:05, Thomas Zimmermann wrote:
There's still the question of the semantics of this parameter. It's a bit fuzzy.
If you use 'disable_handover' (as you mentioned in another mail), it would mean that only the handover itself is disabled. So if simpledrm is not bound to the device, then a native driver should load. That would be hard to implement with the current code base, where we have to take old fbdev drivers into account.
(And to be pedantic, we don't really do a handover of the device. We hot-unplug the generic platform device, so that the driver for the native device can operate the HW without interference.)
Simpledrm only acquires an aperture, but never removes a driver. If there is a driver already, simpledrm would fail. Only native drivers try > to remove drivers and would trigger the test. So your patch is more something like 'disable_native_drivers'.
I'd go with 'disable_native_drivers', or maybe 'disable_device_handover'
That works for me and "drm.disable_native_drivers" is also what Neal meant with his "drm.noplatformdrv", but yours is much easier to remember / type.
I'm good with that too. :)