On Thu, Dec 01, 2016 at 11:20:00PM +0000, ron minnich wrote:
If people are trying native graphics init I still think it's worth trying the SPARK stuff from nico at least once.
I'm intrigued by the use of Ada and excited about applying more formal methods as well as safer languages to the firmware space. It would also be interesting to consider using sel4 or CertiKOS as a payload.
Do we require any sort of VGA init if the payload has the kernel framebuffer and mode support? I should do some tests with that on the x230.
On Thu, Dec 1, 2016 at 2:54 PM Klemens Nanni kl3@posteo.org wrote:
https://item.taobao.com/item.htm?id=43820633263&toSite=main http://forum.51nb.com/thread-1662397-1-1.html
Interesting, I wonder how well these replacement screens would work with coreboot? I assume the proprietary VGA Option ROM won't work with those and native graphics initialisation still needs some work on the X230 either ways.
Based on the auto-translated Chinese forum post, I believe they have tapped into the docking station to get sufficient display lanes. I haven't tried using the docking station with coreboot, so I'm not certain how well it would work either.
Trammell Hudson wrote:
Do we require any sort of VGA init if the payload has the kernel framebuffer and mode support?
As far as I know the answer for more recent kernels is "sortof".
It used to be that i915 in the kernel was completely independent. It would initialize GPU and turn on the panel with native resolution.
At some point in 4.x there was a change in i915 to make it depend on the VBT embedded into the VGA option ROM.
coreboot may or may not be able to generate a satisfactory VBT on its own. In theory of course it could. I just don't expect the C code to be complete.
I like Nico's approach a lot. I would not mind if we integrated that as the graphics init for relevant platforms, looking ahead maybe even expanding its coverage to further platforms.
//Peter
-
Peter Stuge -
Trammell Hudson