Attention is currently required from: Tim Crawford, Jeremy Soller, Tim Wawrzynczak, Michał Kopeć, Michael Niewöhner. Michał Żygowski has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/62496 )
Change subject: drivers/gfx/nvidia: Add Optimus driver based on Intel PCIe RTD3 ......................................................................
Patch Set 3:
(2 comments)
Patchset:
PS1:
So are you folks still interested in updating CB:57034 with something similar to what's in your fork […]
We uploaded this patch totally aware of the other ones lurking on gerrit... The other patches didn't fully work in our case that is why we wanted to push this one as a reference for further work and unify the Optimus support once and for all. The final intent was to abandon the patch when the right implementation lands in.
File src/drivers/gfx/nvidia/optimus/optimus.c:
https://review.coreboot.org/c/coreboot/+/62496/comment/2c3563aa_83a95d46 PS3, Line 195: if (config->enable_gpio.pin_count) {
A GC6 compliant driver must be able to handle the JT DSM methods and decide to skip turning off powe […]
Understood. We didn't implement the GC6 support because we haven't had any specification, expertise and methods to properly implement, yet verify the feature is working as supposed to. Additionally, you mentioned on OSFC that the vendor ACPI code for GC6 is dead, what forced us to leave the driver at the point shown in this patch (ported PCIe RTD3 driver to be as close to vendor ACPI as possible). The advantage is that the patch works under Linux and Windows (although it will simply turn the dGPU on and off as you implicitly said). In the other comment Tim mentioned you got hands on the hardware design guide which let you implement GC6 support. We would like to get rid of the multiplication of the Nvidia driver patches, so I hope we may work together on your Nvidia patchset to get it landed in coreboot tree. We intended to abandon this patch anways, since it is pretty much similar to CB:43615 in terms of ACPI code (back when we developed the Nvidia support, it was still an active patch).