Add LinuxKPI support for attaching DRM drivers

[markj]

Summary

The main difference is DRM drivers attach to a vgapci bus rather than a
pci bus. However, for now we need to override the vgapci ivars with the
grandparent's ivars in order for the rest of the LinuxKPI to work. This
is a hack, but it works fine for now.

Test Plan

Tested with i915 and amdgpu.

[• hselasky]

sys/compat/linuxkpi/common/src/linux_pci.c

137

Is this a more clear way to write the code?

if (pdrv->isdrm) {
   dinfo = device_get_ivars(parent);
   device_set_ivars(dev, dinfo);
} else {
   dinfo = device_get_ivars(dev);
}

267

Is "linux_pci_register_driver_sub()" a better name than "_linux_pci_register_driver()" ?

291

You should add a NULL check here. "pci" might be missing!

302

Same here. Add a NULL check for systems without "vgapci".

[markj]

• Address review comments.

[markj]

sys/compat/linuxkpi/common/src/linux_pci.c

267

I don't feel strongly about this; the "_" prefix seems to be a relatively common way to give a name to an internal wrapper in FreeBSD. If you prefer "_sub" I'll change it.

302

Note that in this case, we create the vgapci devclass if it doesn't exist. In particular, it seems this function won't fail.

[markj]

• Handle devclass_create() failure.

[markj]

sys/compat/linuxkpi/common/src/linux_pci.c

302

Sorry, I skimmed too quickly. We can fail in at least the case where malloc() actually fails to allocate a devclass struct.

[• hselasky]

sys/compat/linuxkpi/common/src/linux_pci.c

267

OK. Either way is fine.

301

This static drm_devclass seems redundant. devclass_add_driver() always assigns the devclass and never reads this one! Can you remove and test?

[markj]

• Get rid of bogus devclass.

sys/compat/linuxkpi/common/src/linux_pci.c

301

Thanks, you're right. Works fine on i915.

[• hselasky]

Looks good. Try to make the commit message a bit clearer:

DRM drivers need to attach to the vgapci bus rather than the PCI bus.
Make own PCI register driver for this purpose.
Further the DRM drivers require the vgapci device has the ivars of
the grandparent for the rest of the LinuxKPI PCI dependencies to
work correctly.