Hello Ron,
Well I've made the patch, added a 13100 specific pci_bus_enable_resources in pciexp_porta.c and modified the pcie_ops structure accordingly. Am not sure this should be enough. Actually these specific functions are never called. pcie_init() part of the initial port is not called as well. I've placed a few printk_debug and I do not see them on the log. And yeah I did buildtarget, make clean, make. Any idea what could go wrong here?
Actually the code in pci_device.c is setting the correct register in the bridges, just the value is incorrect. I am currently checking legacy vs coreboot lspci dumps and investigate differences by differences. It is setting the ISA IO/MEM forwarding bit and we do not want this to happen. If we do that we have to set the other bridges as well. This is related to the 10bit address decoding.
Another thing I did see is that in fact, I do not have VGA working at all. The screen gets correct display at the moment Linux loads the GFX driver. So right now checking the system memory mapping legacy vs coreboot to ensure the required holes are present.
Thank you,
Arnaud
ron minnich wrote:
On Tue, Aug 18, 2009 at 11:04 PM, ron minnichrminnich@gmail.com wrote:
I have this bad feeling that we never converged on answering the question.
Arnaud, did you see that you would want to extend the various
i3100-specific enable_resources
Sorry, left that "i3100" part out.
functions to correctly set up the VGA control bit on your various bridges? Or should we continue this discussion.
ron
* *